Machine Learning Strategies in MicroRNA Research: Bridging Genome to Phenome

Abstract

MicroRNAs (miRNAs) have emerged as a prominent layer of regulation of gene expression. This article offers the salient and current aspects of machine learning (ML) tools and approaches from genome to phenome in miRNA research. First, we underline that the complexity in the analysis of miRNA function ranges from their modes of biogenesis to the target diversity in diverse biological conditions. Therefore, it is imperative to first ascertain the miRNA coding potential of genomes and understand the regulatory mechanisms of their expression. This knowledge enables the efficient classification of miRNA precursors and the identification of their mature forms and respective target genes. Second, and because one miRNA can target multiple mRNAs and vice versa, another challenge is the assessment of the miRNA-mRNA target interaction network. Furthermore, long-noncoding RNA (lncRNA)and circular RNAs (circRNAs) also contribute to this complexity. ML has been used to tackle these challenges at the high-dimensional data level. The present expert review covers more than 100 tools adopting various ML approaches pertaining to, for example, (1) miRNA promoter prediction, (2) precursor classification, (3) mature miRNA prediction, (4) miRNA target prediction, (5) miRNA- lncRNA and miRNA-circRNA interactions, (6) miRNA-mRNA expression profiling, (7) miRNA regulatory module detection, (8) miRNA-disease association, and (9) miRNA essentiality prediction. Taken together, we unpack, critically examine, and highlight the cutting-edge synergy of ML approaches and miRNA research so as to develop a dynamic and microlevel understanding of human health and diseases.

Introduction

In the current post-genomic and multi-omics era, there is growing emphasis on regulation of gene expression. MicroRNAs (miRNAs) are small endogenous noncoding RNAs with a directive role in the posttranscriptional regulation of gene expression across organisms (Ambros et al., 2003; Bartel, 2004; Lagos-Quintana et al., 2001; Lau et al., 2001). The miRNAs are one of the prevalent groups of gene regulatory systems, impacting the expression of protein-coding genes (Lewis et al., 2003).

The biogenesis of miRNA is tightly regulated, and the dysregulation of this process has been linked to both health and susceptibility to various human diseases, including cancers and infections (Peng and Croce, 2016). This article offers the salient and current aspects of machine learning (ML) tools and approaches from genome to phenome in miRNA research.

Overview of miRNA Integrative Biology

The miRNAs are mostly transcribed from intergenic regions and in a promoter-dependent or -independent manner from intronic regions by RNA polymerase II. The biogenesis of miRNAs is not restricted to a canonical pathway; rather, it has also been proposed through multiple noncanonical pathways (Chong et al., 2010).

In the canonical pathway, the transcribed primary miRNA is processed into a 70-nucleotide hairpin loop structure precursor by ribonuclease III (Drosha) (Lee et al., 2003), a component of two multiprotein complexes. The larger complex comprises a diverse array of RNA-associated proteins, including RNA helicases; the smaller complex contains Drosha and the double-stranded heme-binding protein DGCR8 (DiGeorge Syndrome Critical Region 8) (Faller et al., 2007), which is located in the nucleus and its gene deletion characterizes the DiGeorge syndrome (Gregory et al., 2004). Subsequently, the precursor miRNAs are exported from the nucleus to the cytoplasm mostly by the nuclear export receptor XPO5 (Exportin 5) (Yi et al., 2003).

In the cytoplasm, the precursor miRNA undergoes further processing by a second RNAse III-type endonuclease, Dicer, which cleaves the loop structure to liberate an miRNA duplex with a length of 19–22 nucleotides (Cullen, 2004). In the process of miRNA silencing, the methylated mature miRNA duplex separates, and one strand is selectively bound into a protein complex known as the miRNA-induced silencing complex (miRISC), which consists of multiple subunits, including Argonaute (Lagos-Quintana et al., 2001). The other strand is typically degraded. The miRISC engages the complete or partial complementarity binding of miRNA seed region to the 3′-untranslated region (UTR) of the target mRNA, orchestrates mRNA deadenylation, and decapping toward degradation or translational repression (Krol et al., 2010; Siomi and Siomi, 2010). Noncanonical miRNA biogenesis pathways depend on the source of the primary miRNAs, such as miRNAs encoded in the introns (mirtrons), endogenous short-hairpin RNAs, and tRNA fragments.

Together, they contribute to the miRNA pool, as reviewed elsewhere (Ha and Kim, 2014; Kim, 2005).

MiRNAs have the ability to bind to different regions of mRNA, such as the 5′-UTR and protein-coding exon regions, but the base pairing between the miRNA seed region and the 3′-UTR of mRNA is mainly responsible for the quantitative regulation of mRNAs or their protein forms (Lewis et al., 2005). When miRNAs have highly complementary targets, they can induce mRNA degradation accompanied by tailing and trimming. Interestingly, one miRNA can target multiple mRNAs simultaneously, and the reverse is also true. This results in the formation of complex posttranscriptional gene regulatory networks (Enright et al., 2003).

Emerging research has demonstrated the interaction of miRNAs with their targets, including mRNAs (Liu et al., 2014), long-noncoding RNAs (lncRNAs) (Yoon et al., 2013), circular RNAs (circRNAs) (Hansen et al., 2013), and pseudogenes (Thomson and Dinger, 2016). The lncRNAs can regulate the expression and function of miRNAs by acting as miRNA sponges or competing with miRNAs for binding to their target mRNA. The interaction between miRNA on lncRNA and circRNA can be complex and context-dependent, and understanding the specificity or preferentiality of such interactions can provide insights into the dynamicity of regulatory networks. This introduces the necessity for analyzing the layer of the dynamic regulatory network of the miRNAs and their targets in diverse biological conditions (Paraskevopoulou and Hatzigeorgiou, 2016; Yoon et al., 2014).

The influence of miRNAs extends to a multitude of target genes enabling the regulation of the growth and proliferation of cells, tissue differentiation, and embryonic development. In this regard, it is unsurprising that the imbalance in miRNA-mediated regulation has been associated with the emergence of several types of cancer (Hwang and Mendell, 2006). For example, the differential expression of miR-21 results in ovarian, colorectal, and leukemia cancers (Adams et al., 2014). Identifying and interpreting the functional impact of miRNAs using experimental approaches are challenging, given the complexity of their regulatory target networks in specific biological conditions.

These challenges extend to the analysis of the miRNA coding potential of genomes, prediction of their promoters, characterization of precursor miRNAs, identification of mature miRNA sequences, their potential interactions with noncoding RNAs, analysis of their mRNA target specificity and diversity, and subsequently, the functional characterization of miRNAs in specific biological processes and their association with diseases/disorders (Lauria et al., 2023; Megret et al., 2022). The initial addressing of miRNA for better identification starts from its nomenclature. In miRBase, miRNA nomenclature infers “hsa” as an abbreviation for Homo sapiens, following scientific conventions. This system, based on the initial letters of “Homo” and “sapiens” (h from Homo and sa from sapiens), ensures a standardized and concise representation for human miRNA nomenclature (Griffiths-Jones et al., 2006).

Owing to the limitations in the experimental analysis of miRNA biology increases the essentiality of computational dynamics. Indeed, in the past decade, different computational approaches have been undertaken for unraveling the heterogeneity of miRNAs and their associated molecular pathways. In the past, ML approaches have been used to analyze miRNA turnover, miRNA-target interactions, disease associations, characterize pre- and mature miRNAs, and predict miRNA promoters. Availability of curated experimentally validated miRNAs, their targets, and their association with diseases facilitated by public databases such as miRbase (Kozomara et al., 2019) miRTarBase (Hsu et al., 2011), miRecords (Xiao et al., 2009), Human miRNA disease database (HMDD) (Huang et al., 2019b), and miR2disease (Jiang et al., 2009) has accelerated the progressive development of ML tools.

Most ML-based mRNA target prediction tools, such as TarPmiR (Ding et al., 2016), miRaw (Pla et al., 2018), and miTar (Gu et al., 2021), are designed to use tens to hundreds of different features in their models, such as the sequence, structure, conservation, thermodynamic properties, and context of the interacting molecules. Recently, computationally predicted data architecture has also been accounted for by unsupervised, supervised, and reinforcement learning methods on miRNA biology (Azari et al., 2023; Pawelka et al., 2022).

The availability of new features and data sets of high-throughput information on miRNA-target and other molecular interactions has led to the development of enhanced ML-based methods, such as miES (Song et al., 2019) and miRcorrNet (Yousef et al., 2021). Such tools explore miRNA's involvement in well-known signaling pathways and their associations with diseases. Mining these tools helps to unpack the multilevel prioritization of miRNA. Most tools are made in isolation and must be assembled for effective comparative analysis and visualization by the scientific community to accelerate miRNA-related applications.

The comprehensive information in the following expert analyses offers an extensive compendium of tools and approaches to miRNA, augmented by ML techniques, along with their meticulously stringent comparative evaluation. In structuring our review, we aimed to provide a comprehensive yet focused examination of ML approaches in miRNA research, spanning both the nuclear and cytoplasmic levels of miRNA processing. Our selection of listed items within each topic was guided by the intention to highlight the biological process of miRNA and its interconnected process and how the implementation of these biological factors in an ML perspective. This insight addresses the foundational steps in miRNA biogenesis and regulation, laying the groundwork for subsequent cytoplasmic processing and functional analysis. Thus, our review represents a snapshot of current advancements rather than an exhaustive catalog of all the available approaches.

Genomic and Nuclear-Level Features of miRNA

miRNA prediction in genome

Detecting novel miRNA genes poses a significant challenge in bioinformatics, requiring the integration of various factors such as the characteristic secondary structure of the precursor, the conservation of both primary and precursor sequences, and expression data encoded in small RNA libraries (Hertel et al., 2014).

The search for novel or species-specific miRNAs involves scanning the entire genome or genome-wide alignments with related species using ML approaches. These methods consider various features of the precursor, including its fold-back structure and conservation information, to classify candidate sequences. ML algorithms are trained on sets of known miRNAs especially from the miRBase database (Griffiths-Jones et al., 2006), often incorporating data sets representing sequences with similar characteristics but are not miRNAs. Computational-based tools exist for this purpose, such as miRCheck (Jones-Rhoades, 2010), MapMi (Guerra-Assuncao and Enright, 2010), and miROrtho (Gerlach et al., 2009) differing in their hairpin detection algorithms, feature definitions, and ML approaches. While some tools focus on identifying precursor structures, others prioritize the detection of mature miRNAs by analyzing precursor hairpins and mature miRNA candidates, along with flanking sequence information.

The miRNAs are mostly transcribed by RNA polymerase II and posttranscriptionally modified mainly by 5′capping, splicing, and polyadenylation at the 3′end, to become mature and functional miRNAs (Cai et al., 2004; Lee et al., 2004). For intronic, exonic, or intergenic miRNAs, at the transcription initiation sites, the physical association of polymerase with the promoter are facilitated by transcription factors such as p53 (Tarasov et al., 2007), MYC (Chang et al., 2008b), ZEB1, and ZEB2 (Bracken et al., 2008) and epigenetic regulators (Bueno et al., 2008; Kim et al., 2009). DNA methylation and histone modifications contribute to the regulation of miRNA gene expression (Davis-Dusenbery and Hata, 2010; Lujambio et al., 2008; Scott et al., 2006).

These regulative features, such as histone modification and nucleosome positioning patterns, are used for the computational prediction of miRNA biogenesis (Ozsolak et al., 2008). In this context of identification of polyadenylated miRNA, transcripts have been identified as expressed sequence tags (ESTs) (Chien et al., 2011). Considerable work has been put into developing computational algorithms that integrate chromatin immunoprecipitation (ChIP)-seq data, deoxyribonuclease sequencing, and transcriptome profiles to decode transcriptional-level miRNA regulations (Hollbacher et al., 2020).

The primary miRNA hairpin contains a long imperfect stem of ∼30 bp with a single-/double-stranded junction (Han et al., 2006). Multiple mismatches and wobble base pairs in the upper stem of primary miRNA are considered to analyze the efficiency and accuracy of their processing (Li et al., 2020b). The characterization of precursor miRNA to 60–120 nucleotides by Drosha and DGCR8 accounted for sequence length, number of base pairings, and the free energy for the secondary structure (Starega-Roslan et al., 2011). The unstable nature of the secondary structure depicts that precursor miRNA may be degraded before it can be processed (Suzuki et al., 2011). The presence of stabilizing sequences such as GC regions or hairpin loops and RNA binding proteins necessitates stability (Bail et al., 2010). Figure 1 shows a schematic representation of the miRNA biogenesis in the nuclear level and its associated tools and approaches.

FIG. 1.

Machine learning-based tools and approaches to processing microRNA biogenesis at the nuclear level. From transcription to precursor microRNA level. Tools/approaches and corresponding processes are also included herein. The year of publication is given in brackets. Some approaches do not have names, so we refer to them with the attendant authors' names.

miRNA promoter prediction

The promoters are essential regions accounted for by different regulatory factors to regulate transcription mostly by RNA polymerase II. Mapping the precise locations of the miRNA promoter regions is crucial for revealing the transcriptional regulatory interface (Weis and Reinberg, 1992; Zeng et al., 2009). Transcription factor-miRNA regulation databases such as TransmiR v2.0 (Tong et al., 2019), microPIR (Piriyapongsa et al., 2012), and miRT (Bhattacharyya et al., 2012a) facilitate the data set collection. The traditional prediction methods have used several genome sequence features, including collective analysis of ESTs and different genome sequence patterns such as TATA box, CAAT box, and GC box (Abeel et al., 2008; Bajic et al., 2004; Ohler and Niemann, 2001).

The culmination of modern technologies results in ML tools; the primary one common query voting (CoVote) used nucleotide sequence motif features in terms of k-mers by building a discriminative model to distinguish polymerase II and polymerase III promoters. Finding the possible k-mers with the help of WordSpy motif-finding algorithm analysis of sequence motifs based on base-pair length results in a support vector machine (SVM)-based classifier (Zhou et al., 2007).

Regarding text data vectorization, the evolution of word embedding methods has considerable untapped potential in addressing sequence-related issues. Similar kinds of traditional nucleotide sequence formats and the addition of CpG island features have been incorporated in another SVM-based approach developed by Bhattacharyya et al. (2012b). Subsequently, CoreBoost_HM, a boosting algorithm approach developed in 2009, used more specific properties regarding gene-regulation epigenetic control markers such as histone markers, acetylation, and methylation signatures and previously adopted CpG island features instead of k-mers (Wang et al., 2009b).

Five years later, the SVM-based miRNA transcription start site prediction tool further integrated high-resolution RNA-sequencing data with active transcription sites derived from ChIP-seq and deoxyribonuclease sequencing, which has revealed dicistronic and polycistronic miRNA transcripts (Georgakilas et al., 2014).

Recently, D-miRT, a neural network-based tool, contributed high-resolution features on the basis of single-base resolution of nucleotides (Cha et al., 2021). In addition, there has been a surprising fact that natural language processing algorithms specifically bidirectional encoder representations from transformers (BERT) lead to a new approach called miProBERT (Wang et al., 2023). The model utilizes the pretrained DNABERT and fine-tunes it on the gene promoter data set to include information about the biological properties of promoter sequences. The model is then used to scan the upstream regions of intergenic miRNAs, resulting in the identification of 665 miRNA promoters. The model's discriminative ability is improved by using a random substitution strategy to construct a negative data set, which reduces the false-positive rate to 0.0421.

On independent data sets, miProBERT outperforms other gene promoter prediction methods and is shown to have a precision of 78.13% and recall of 75.76%. An overview of these algorithms or approaches is summarized in Table 1.

Table 1.

MicroRNA-Promoter Prediction Tools

SI. No	Tool name	Algorithm	Reference	Salient features
1	CoVote (2007)	Support vector machine and wordspy	Zhou et al. (2007)	Sequence (K-mers)
2	CoreBoost_HM (2008)	Boosting algorithm	Wang et al. (2009b)	Histone modification profile and DNA sequence features
3	miRStart (2011)	Support vector machine	Chien et al. (2011)	Sequence conservation, CpG density, histone modifications, TATA box
4	Bhattacharyya et al. (2012)	Support vector machine	Bhattacharyya et al. (2012b)	N-mer, palindromic, special subsequence pattern, CpG island features
5	microTSS (2014)	Support vector machine	Georgakilas et al. (2014)	ChIP-seq and deoxyribonuclease sequencing features
6	D-miRT (2021)	Convolutional neural network	Cha et al. (2021)	DNase-Seq and histone modification, nucleotide resolution
7	miProBERT (2023)	BERT	Wang et al. (2023)	GC content, CpG density, TATA and CCAAT boxes, sequence conservation, minimum free energy, hairpin loops

BERT, bidirectional encoder representations from transformers; ChIP, chromatin immunoprecipitation; CoVote, common query voting; microTSS, miRNA transcription start site.

miRNA precursor classification

Intermediary precursor miRNA and its hairpin loop structures have a major importance in miRNA biogenesis. Computational techniques for miRNA prediction use comparative genomic methodologies such as homology-based searches (Berezikov et al., 2005), genome-wide scanning (Yones et al., 2018), phylogenetic footprinting (Gao et al., 2013), and clustering analysis (Sewer et al., 2005) to find potential precursor miRNAs from candidate hairpins. Classification or prediction of the precursor miRNAs is challenging as many sequences (over 11 million estimated) can fold into precursor miRNA-like hairpin secondary structures (Bentwich et al., 2005). Many of these could be predictably pseudohairpins and require efficient classification. The exponential growth of databases such as miRGen (Megraw et al., 2007) and miRCarta (Backes et al., 2018) has contributed to data sets for the training and testing aspects in ML methods.

The initial methods for classifying the precursor miRNAs account for distant interactions between nucleotides of known genomic sequences. Among these methods, the basic ML hidden Markov model-based tool ProMiR (Sewer et al., 2005) and the context-sensitive hidden Markov model (CSHMM) (Agarwal et al., 2010) use free energy, structural and sequential in terms of the stem's length, number, and proportion of nucleotides. The hidden Markov model is a formal foundation for making probabilistic models of linear sequence labeling problems. During the initial time HMM serves for a diverse range of problems, including genefinding, profile searches, multiple sequence alignment, and regulatory site identification. Most classification models are built by the supervised learning model SVM, which includes Triplet-SVM (Xue et al., 2005), RNAmicro (Hertel and Stadler, 2006), miPred (Jiang et al., 2007), microPred (Batuwita and Palade, 2009), MiRenSVM (Ding et al., 2010), miRNA-dis (Liu et al., 2015), and miRBoost (Tran Vdu et al., 2015).

The mentioned classification models are all utilized by finding the hyperplane that best separated the input data into different classes. The deployment of these tools is based on the information of structured sequence by analyzing the continuously paired or unpaired status of the nucleotide of precursor miRNA and pseudo-pre-miRNA.

The continuously paired nucleotides are part of a stem structure consisting of base pairs that are complementary to each other and form a double-stranded region of the precursor miRNA molecule. In contrast, unpaired nucleotides are part of a loop structure that forms a single-stranded region of the precursor miRNA molecule. The accuracy of these models varies, but they have generally shown 85% above in prediction tasks. Incorporating similar features such as the minimum free energy, local contiguous triplet structure composition, and dinucleotide shuffling results in random forest methods MiPred (Jiang et al., 2007) and HuntMi (Gudys et al., 2013).

Random forest works by combining a multitude of decision trees at training time and outputting the class by the mean prediction of the individual trees. The MiPred can classify precursor miRNA from other noncoding RNAs and protein-coding mRNAs with an accuracy of 92.5 on a set of experimentally validated targets, while HuntMi is designed to identify precursor miRNA from genome sequences by incorporating additional features such as sequence motifs and evolutionary conservation parameters.

Subsequent advancement in the neural networks further contributes many tools in the field of miRNA precursor classification resulting in the development of various tools such as MiRANN (Rahman et al., 2012), deepMiRGene (Seunghyun et al., 2016), DP-miRNA (Tasdelen and Sen, 2021; Thomas et al., 2017; Zheng et al., 2019b), miRNAFinder (Lokuge et al., 2022), and miRe2e (Raad et al., 2022). The accuracy of these tools can vary depending on the data set and task they were evaluated on. Sequential characteristics of the MiRANN enclose the frequency of dinucleotide pairs, the percentage and the propensity of nucleotide bases, and the structure encompassing the number and length of stem-loops and hairpins.

The deepMiRGene approach learns significant characteristics from data without manual feature engineering and produces a model that can appropriately depict the structural properties of precursor miRNAs with the help of long short-term memory (LSTM) networks. The innovative classification methodology put forth by Zheng et al. (2019b) and Tasdelen and Sen (2021) is predicated upon the combined architecture of convolutional and neural networks. As a corollary, they have utilized the one-hot encoding as a means of text data vectorization. Overall, these tools effectively represent a paradigm shift to identify the structural and thermodynamic features of precursor miRNAs, contributing to the ongoing research in this field. An overview of these algorithms or approaches is summarized in Table 2.

Table 2.

MicroRNA Precursor Classification Tools

SI. No	Tool name	Algorithm	Reference	Salient features
1	ProMiR (2005)	Hidden Markov model	Sewer et al. (2005)	Length and distance of hairpin loop, number of nucleotides in the loop, the proportion of nucleotides
2	Triplet-SVM (2005)	Support vector machine	Xue et al. (2005)	Triplet structure sequence elements
3	RNAmicro (2006)	Support vector machine	Hertel and Stadler (2006)	Structure, sequence composition, sequence conservation, thermodynamic stability, structure conservation
4	MiRFinder (2007)	Support vector machine	Huang et al. (2007)	Mutation profile of miRNA
5	MiPred (2007)	Random forest	Jiang et al. (2007)	Local contiguous triplet structure, free energy
6	mir-KDE (2008)	Kernel density estimator	Chang et al. (2008a)	Sequence, base pairing propensity, minimum free energy, base pair distance, degree of compactness
7	MicroPred (2009)	Support vector machine	Batuwita and Palade (2009)	Sequence, structure, free energy
8	CSHMM (2010)	Hidden Markov model	Agarwal et al. (2010)	Secondary structure
9	MiRenSVM (2010)	Support vector machine	Ding et al. (2010)	Sequence, structure, thermodynamics, base pair
10	MiRANN (2012)	Artificial neural network	Rahman et al. (2012)	Sequence, structure, energy, G + C content
11	HuntMi (2013)	Random forest	Gudys et al. (2013)	Sequence, structure, free energy
12	miRNA-dis (2015)	Support vector machine	Liu et al. (2015)	Sequence, structure
13	miRBoost (2015)	Support vector machine	Tran Vdu et al. (2015)	Sequence, structure, energy
14	deepMiRGene (2016)	RNN, LSTM	Seunghyun et al. (2016)	Sequence, structure
15	DP-miRNA (2017)	Deep neural network	Thomas et al. (2017)	Sequence, structure, energy, hairpin length, loop length, base-pair
16	Zheng et al. (2019)	Convolutional neural network	Zheng et al. (2019b)	Length and base composition, secondary structure
17	Abdulkadir Tasdelen et al. (2021)	CNN, LSTM	Tasdelen and Sen (2021)	Sequence and structure
18	miRNAFinder (2022)	Multilayer perceptron	Lokuge et al. (2022)	Sequence, structure, thermodynamic
19	miRe2e (2021)	Neural network attention mechanism	Raad et al. (2022)	Sequence

CNN, convolutional neural network; CSHMM, context-sensitive hidden Markov model; LSTM, long short-term memory; miRNA, microRNA; RNN, recurrent neural network.

Cytoplasmic-Level Features of miRNA

The nuclear export of precursor miRNA is facilitated by karyopherin exportin 5 in the presence of the Ran-GTP cofactor (Yi et al., 2003). The inhibition of exportin 5 is responsible for downregulated expression and increased nuclear localization of the ribonuclease III superfamily of bidentate nucleases, the dicer, which is required to generate an miRNA duplex (Zeng and Cullen, 2004). Subsequent selective loading of AGO protein results in an effector complex called RNA-induced silencing complex (Okamura et al., 2004). Following miRNA duplex loading, the RISC containing Argonaute (Lagos-Quintana et al., 2001) proteins undergoes the crucial process of passenger strand removal. This intricate mechanism is facilitated by the combined action of endonuclease C3PO and the multimeric complex of translin and its associated protein X (Liu et al., 2009).

Unwinding of the miRNA duplex is a process resulting because of the mismatches at nucleotide positions 2–8 and 12–15 in the guide strand (Kawamata et al., 2009). miRNAs bind to AGO with their 5′end embedded inside the AGO MID domain and the 3′end docked at the AGO PAZ domain (Ma et al., 2005; Schirle and MacRae, 2012). Through partial base pairing, miRNA guides the RISC to target mRNAs (Yang et al., 2020a). The mature miRNA identification by ML parameters accounts for stem-loop structure, conserved terminal nucleotides, seed region, mismatches, and thermodynamic stability (Rojas et al., 2020; Wang et al., 2005). The lower stem-loop contains the cleavage site for the dicer enzyme, a key feature for localizing the mature miRNA (Han et al., 2006; Sacar et al., 2013).

Highly stable miRNA duplexes may be less accessible to the RNA-induced silencing complex and, therefore, less effective in regulating gene expression (Hibio et al., 2012). MiRNA biogenesis in the cytoplasmic level and the associated tools and approaches are represented in Figure 2.

FIG. 2.

Machine learning-based tools and approaches on microRNA biogenesis at the cytoplasmic level processing. From precursor microRNA to various interactions of microRNA with mRNA, lncRNA, and circRNA. MicroRNA disease association prediction tools and approaches are also included herein. circRNA, circular RNA; lncRNA, long noncoding RNA.

Mature miRNA prediction

Further exploration of miRNA envisages the functional characteristics of mature miRNA. Sequential and structural assessment of mature miRNA using ML revolves around miRNA duplex aspects in terms of base pair counts and bulges on the duplex, and frequency of nucleotides. Because there can be deviations from perfect base pairing in the duplex, resulting in bulges or mismatches. Bulges occur when one strand has an unpaired nucleotide, creating a loop in the duplex. Mismatches occur when two nucleotides on opposing strands do not form a perfect Watson crick base pair. These popular conceptions result in the Bayesian framework, SVM, and random forest-based approaches, listed in Table 3.

Table 3.

Mature MicroRNA Prediction Tools

SI. No	Tool name	Algorithm	Reference	Salient features
1	MatureBayes (2010)	Naive Bayes	Gkirtzou et al. (2010)	Sequence, structure, position
2	Maturepred (2011)	Support vector machine	Xuan et al. (2011)	Structure, stability, energy, position-specific
3	MiRmat (2012)	Random forest	He et al. (2012)	Sequence, structure, energy
4	Mirdup (2013)	Random forest	Leclercq et al. (2013)	Sequence, structure, energy
5	miRLocator (2015)	Random forest	Cui et al. (2015)	Sequence, structure, energy
6	MiRduplexSVM (2015)	Support vector machine	Karathanasis et al. (2015)	Sequence, structure, thermodynamic
7	Matpred (2015)	Radial basis function and support vector machine	Li et al. (2015)	Position-specific structure, free energy, distances to the stem-loop

For example, in the naive Bayes approach MatureBayes, the mature miRNA is represented as a two-dimensional character array containing information about the base composition and structure regarding match and mismatch (Gkirtzou et al., 2010), while in MiRduplexSVM, the nucleotide bases are represented by four binary variables that has an accuracy of 85.4% (Karathanasis et al., 2015). MatureBayes achieved an accuracy of 94.5 on a benchmark data set of human miRNAs, which was comparable with the best performing miRNA prediction tools at the time. The statistical ML approach Bayesian framework allows to quantify uncertainty in the estimates and predictions by using the probability distributions instead of point estimates. Another tool, MatPred, identified mature miRNA within novel pre-miRNAs by considering the duplex window region, lower stem-loop, and minimum free energy. Around 94 features have been extracted from the mature miRNA loci and flanking regions.

The decision function of the MatPred is the SVM with the radial basis function kernel (Li et al., 2015). On the contrary, MiRmat has features such as the size and length of internal loops and the number of bulges in mature sequence regions for dicer site prediction (He et al., 2012). The Maturepred includes position-specific features, stability-related features, minimum free energy, and local contiguous triplet structure (Xuan et al., 2011).

miRNA target prediction

Comprehending the overarching picture of miRNA is sustained through rigorous investigations of the most extensively studied partial complementary interaction with mRNA (Parveen et al., 2019; Piletic and Kunej, 2018). The miRNA recognizes its targets by incomplete base pairing to sequence motifs most often present in the 3′-UTR of their target mRNAs (Bartel, 2009). This partial complementarity inflates the count and quality of potential targets and the development of ML tools (Chen et al., 2019a; Reyes-Herrera and Ficarra, 2012). Computational approaches commence with sequence complementarity regarding seed region (Krek et al., 2005) and thermodynamic potentials in terms of stability of miRNA binding to the target mRNA, which takes the measure as Gibbs free energy (Mathews et al., 1999) and evolutionary conservation such as the keeping of sequence across species (Peterson et al., 2014).

The progressive evolution of manually curated database of experimentally validated miRNA events such as miRbase (Kozomara et al., 2019), MiREDiBase (Marceca et al., 2021), and miRDB (Chen and Wang, 2020) currently explodes the target interaction prediction. The classical RNA secondary structure prediction algorithm extends to dynamic programming-based RNA hybrid that uses minimum free energy hybridization as a parameter for target prediction (Rehmsmeier et al., 2004). Instead of only looking at the free energy score and binding in the 5′score, boosted genetic programming-based TargetBoost has discovered a pattern in the miRNA mRNA binding characteristics (Saetrom et al., 2005). The limitations in the reliability of these approaches for feature selection and data coverage led to the implementation of different models such as SVM, naive Bayes, neural networks, and random forest-based ML approaches, listed in Table 4, because of their analytical capability of nonlinear connections.

Table 4.

MicroRNA Target Prediction Tools

SI. No	Tool name	Algorithm	Reference	Salient features
1	RNAhybrid (2004)	Extension Algebraic Dynamic Programming	Rehmsmeier et al. (2004)	Thermodynamic
2	TargetBoost (2005)	Genetic Programming	Saetrom et al. (2005)	Position-based
3	Kim et al. (2006)	Support vector machine Classifier Kernel method	Kim et al. (2005)	Position-based, structural thermodynamic
4	MiTarget (2006)	Support vector machine	Kim et al. (2006)	2,3,4,5,7,6, Position, Free energy, mismatches, matches (AU, GU) at the 5′ part and 3′ (15 features)
5	Yan et al. (2007)	Adaboost with several support vector machine classifiers	Yan et al. (2007)	Target site, seed region, nonseed region of the target site and mRNA local secondary structure
6	NbmirTar (2007)	Naive Bayes	Yousef et al. (2007)	Seed and out seed sequence, duplex structure
7	MirTarget2 (predicted targets imported to mirDB) (2008)	Support vector machine	Wang and El Naqa (2008)	Sequence and structure
8	Liu et al. (2008)	Support vector machine	Liu et al. (2008)	Position specified and regional features
9	MirTif (2008)	Support vector machine	Yang et al. (2008)	Seed nonseed matches, mismatches
10	Targetminer (2009)	Support vector machine	Bandyopadhyay and Mitra (2009)	Seed matching site, miRNA-mRNA base interaction in seed region, site accessibility
11	TargetSpy (2010)	Multiboost classifier (support vector machine, Naive Bayes, C4.5, Adaboost)	Sturm et al. (2010)	Number of bulges, site accessibility, number of base pairings
12	MTar (2010)	Artificial Neural network	Chandra et al. (2010)	Mismatches, wobble pair, free energy, hybridization, positional score, seed score, out seed score
13	mirSOM (2011)	SOM	Heikkinen et al. (2011)	Sequence, structure, thermodynamic
14	Homotarget (2013)	Pattern recognition neural network and principal component analysis	Ahmadi et al. (2013)	Seed score, WC pairs, Wobble pairs, mismatches, number of bulges, free energy, base proportions
15	RFmirTarget (2013)	Random forest	Mendoza et al. (2013)	Sequence, free energy, mismatches, seed region, structure
16	mirMark (2014)	Gaussian support vector machine	Menor et al. (2014)	Seed match, structure, free energy
17	miRTDL (2016)	CNN	Shuang et al. (2016)	Match score, mismatch, seed score, free energy, site accessibility
18	Tarpmir (2016)	Random forest	Ding et al. (2016)	Seed match, site accessibility, folding energy, stem and flanking conservation, positional features
19	MiRNATIP (2016)	Self-Organizing Map (SOM)	Fiannaca et al. (2016)	Seed pairing, the presence of bulges, free energy constraints for stability of the duplex
20	DeepMirtar (2018)	Stacked autoencoder	Wen et al. (2018)	Seed match, sequence composition, energy, site accessibility, conservation, distance, hot encoding
21	miRaw (2018)	Feed-forward ANN	Pla et al. (2018)	Thermodynamic structure and sequence
22	miTar (2021)	Hybrid deep learning	Gu et al. (2021)	Structure and sequence

SOM, self-organizing maps.

Considerable accurate output accounts position based, structural, and thermodynamic, and it incorporates single-point mutation, seed pairing, nonseed region secondary structure, conservation, free energy, site accessibility, target site abundance, local AU content, GU wobble in the seed match, and seed pairing stability (Lee and Shin, 2012). The first two to eight nucleotides starting at the 5′end and counting toward the 3′end constitute the seed sequence; either side of the seed sequence refers to the flank sequence (Chipman and Pasquinelli, 2019). For example, the seed properties in terms of the number of bulges, symmetric and asymmetric loops are illustrated as feature sets in the naive Bayes-based NbmiRTar tool that does not rely on conserved sequence features (Yousef et al., 2007).

Data sets having reliable features contribute precision to the output, the SVM-based Targetminer uses a multistage filtering strategy that extracts 90 targeted site context attributes (Bandyopadhyay and Mitra, 2009). SVM is a supervised ML algorithm that works by creating a hyperplane that separates data into different classes based on their features (Noble, 2006). In this context of miRNA target prediction, SVM is trained on a set of known miRNA-target interactions to identify new correlations. One advantage of SVM is that it can also be trained with different kernel functions, such as linear, polynomial, and radial basis functions, to capture nonlinear relationships between miRNA and mRNA sequences. For example, tools such as miTarget (Kim et al., 2006), MirTarget (Wang and El Naqa, 2008), MirTif (Yang et al., 2008), and mirMark (Menor et al., 2014) functioned on this.

The advancements have enabled researchers to design and develop more complex models with deeper layers and improved training algorithms, leading to enhanced performance and efficiency. As a result, these state-of-the-art models such as Mtar (Chandra et al., 2010), Homotarget (Ahmadi et al., 2013), miRNATIP (Fiannaca et al., 2016), miRTDL (Shuang et al., 2016), Deep MirTar (Wen et al., 2018), and miRaw (Pla et al., 2018) interpret the experimentally verified miRNAs and their targets by analyzing their binding site-specific interactions on seed region. The combinatorial functioning of RNN (recurrent neural network) and CNN (convolutional neural network) results in hybrid deep learning approaches miTar (Gu et al., 2021). The CNNs are utilized to extract features from the miRNA-mRNA duplex specifically the base paring and duplex stability, while LSTMs are designed to model temporal dependencies in sequential data, making them well suited for analyzing time-series data. Nowadays, the LSTM-based architectures are demonstrating miRNA target detection (Talukder et al., 2022).

miRNA-lncRNA interaction prediction

Evidence accumulated over the past decade shows that regulation of gene expression involves the interaction with miRNA and lncRNA. The lncRNA biogenesis led to the modulation of chromatin function (Chu et al., 2011; Isoda et al., 2017; Thatai et al., 2023), cytoplasmic mRNA stability and translation (Cesana et al., 2011), and interfere with the signaling pathway (Zhao et al., 2021). In addition, many lncRNAs bearing miRNA complementary sites for miRNAs enabling them to act as competitive endogenous RNAs or sponges of miRNAs that reduce the availability to the miRNA-mRNA target interaction. The stoichiometric relationship between a potential competitive endogenous lncRNA and miRNA is important for achieving a measurable effect on target-mRNA expression (Statello et al., 2021). For example, the lncRNA H19 is found to modulate let-7 family of miRNAs (Kallen et al., 2013).

Several studies have begun to uncover the interactions between miRNA and lncRNA, and more details about the influence of miRNA on lncRNA function are coming into view. Despite extensive research by the exposure of miRNA-lncRNA databases DIANA-LncBase v3 (Karagkouni et al., 2020), LncCeRBase (Pian et al., 2018), and LncRNADisease 2.0 (Bao et al., 2019), the ability to predict miRNA-lncRNA interactions using computational algorithms still needs to be improved by the complexity of the abundance of false-positive results. Resolving the false results leads to several ML tools, listed in Table 5.

Table 5.

MicroRNA-Long Noncoding RNA Interaction Target Prediction Tools

Sl. No	Tool name	Algorithm	Reference	Salient features
1	EPLMI (2018)	Bipartite graph	Huang et al. (2018a)	LncRNA/miRNA similarity matrix
2	GBCF (2018)	Bayesian collaborative filtering model	Huang et al. (2018b)	Sequence-based similarity of RNA, expression profile based, function based
3	SLNPM (2019)	Linear neighborhood similarity measure	Zhang et al. (2019b)	LncRNA-lncRNA similarity, miRNA-miRNA similarity, lncRNA-miRNA interactions
4	GCLMI (2019)	Graph convolution autoencoder	Huang et al. (2019a)	LncRNA-miRNA interaction network
5	LncMirNet (2020)	Deep convolutional neural network	Yang et al. (2020b)	RNA sequence k-mer features, composition/transition/distribution features
6	PmliPred (2020)	CNN and gated recurrent unit	Kang et al. (2020)	K-mer frequency, GC content, number of base pairs, minimum free energy
7	LMI-Dforest (2020)	Deep forest and autoencoder	Wang et al. (2020)	LncRNA-lncRNA similarity, miRNA-miRNA similarity, lncRNA-miRNA interactions
8	LMNLMI (2020)	Multimodal networks	Hu et al. (2020)	LncRNA-lncRNA similarity, miRNA-miRNA similarity, lncRNA-miRNA interactions
9	CIRNN (2020)	CNN and IndRNN	Zhang et al. (2020)	Sequence-based similarity
10	SPMLMI (2021)	Structural perturbation method	Xu et al. (2021)	LncRNA-lncRNA similarity, miRNA-miRNA similarity, lncRNA-miRNA interactions
11	LMI-INGI (2021)	Graphlet interaction	Xu et al. (2021)	Sequence-based similarity
12	PmliPEMG (2021)	Greedy fuzzy decision	Kang et al. (2021)	K-mer frequency, sequence, and structure composition
13	MD-MLI (2022)	Recurrent neural network	Song et al. (2022)	K-mer frequency, GC content, number of base pairs, minimum free energy
14	GKLOMLI (2022)	Linear optimization algorithm	Wong et al. (2023)	LncRNA-lncRNA similarity, miRNA-miRNA similarity, lncRNA-miRNA interactions
15	MILNP (2022)	Linear neighborhood similarity measure	Cai et al. (2022)	K-mer frequency, GC content, number of base pairs, minimum free energy
16	PreMLI (2022)	Deep learning	Yu et al. (2022b)	Sequence similarity (rna2vec)
17	BoT-Net (2022)	LSTM	Asim et al. (2022)	Sequence similarity
18	GEEL-FI (2022)	Deep attention neural network	Zhao et al. (2020)	LncRNA-lncRNA similarity, miRNA-miRNA similarity, lncRNA-miRNA interactions
19	GCNCRF (2022)	Graph convolution network	Wang et al. (2022a)	LncRNA-lncRNA similarity, miRNA-miRNA similarity, lncRNA-miRNA interactions

lncRNA, long noncoding RNA.

The upscale growth of neural networks achieved a great success rate in the miRNA lncRNA prediction. A substantial increase of addressing the common methodology, sequence similarity matrix involves incorporation of various kinds of neural network layers for miRNA-lncRNA prediction. The complementary binding strategy of miRNA on lncRNA reflects the sequence similarity matrix as a popular conception in the computational aspects that relies on the seed region. Conservation of lncRNA sequences between different species suggests that they may have important functional roles, and identifying conserved miRNA binding sites within these sequences can provide further evidence for their functional relevance.

A deep convolutional neural network approach was called lncMirNet (Yang et al., 2020b), witnessing k-mer RNA sequence and distributed representation feature of RNA sequence by doc2vec with an accuracy of 85%. The ability of this technique has been adapted to represent RNA sequences as well where the sequence acts as a document and the individual context-dependent meanings of nucleotides or k-mers are treated as words. For SLNPM (Zhang et al., 2019b), sequence similarity and interaction profile similarity by giving mathematical notations for lncRNA interaction profile, lncRNA sequence similarity, and lncRNA interaction profile similarity (k-mer feature). It shows a 99% accuracy.

Bayesian collaborative filtering model called GBCF to pick up ranking list for an individual miRNA or lncRNA based on the known miRNA-lncRNA interactions in the lncRNASNP (Gong et al., 2015) database and it gives sequence, expression, and biological function-based similarity level. The LMI-INGI (Zhang et al., 2021) interactome network and graphlet interaction by sequence similarity matrix regularized the Smith–Waterman algorithm. GKLOLMI link prediction model for inferring miRNA-lncRNA interactions by Gaussian kernel-based method on network profile and linear optimization algorithm by constructing RNA sequence similarity.

miRNA-circRNA interaction prediction

circRNAs, a vast category of noncoding RNAs that are generated through a noncanonical splicing mechanism called backsplicing (Kristensen et al., 2019; Zhang et al., 2013). The evolutionary evidences say that more than 40 years ago, circRNA molecules were discovered in Viroids (Sanger et al., 1976). A few years later, circRNAs were observed in the cytoplasmic fractions of eukaryotic cell lines by electron microscopy (Hsu and Coca-Prados, 1979). Recent studies have highlighted the importance of circRNAs in a variety of biological processes, including development, aging, and disease. Some circRNAs are highly abundant and evolutionarily conserved, suggesting that they may play critical roles in fundamental cellular processes. Furthermore, circRNAs can act as miRNA or protein inhibitors, regulate protein function, or even be translated themselves, highlighting their potential as targets for therapeutic intervention (Memczak et al., 2013; Xia et al., 2017).

The emergence of circBase (Glazar et al., 2014), circRNADb (Chen et al., 2016a), Circad (Rophina et al., 2020), CIRCpedia v2 (Dong et al., 2018), and CircR2Cancer (Lan et al., 2020) databases put forth a standardized information on circRNA. This leads to the information on the interactive association of circRNA with miRNA, and further discovers a biological scenario, listed in Table 6. The construction of circRNA-miRNA interaction matrix on the basis of similarity and associations describes the methodological part of the ML approaches. Most of the recent approaches are built on the neural network-based methods such as WSCD (Guo et al., 2022), CMIVGSD (Qian et al., 2021), GCNCMI (He et al., 2022), SGCNCMI (Yu et al., 2022a), KGDCMI (Wang et al., 2022b), and NGCICM (Ma et al., 2023). These follow an interaction profile kernel similarity network association of circRNA-miRNA.

Table 6.

MicroRNA-Circular RNA Interaction Target Prediction Tools

Sl. No	Tool name	Algorithm	Reference	Salient features
1	NECMA (2021)	Network embedding method	Lan et al. (2021)	circRNA and miRNA similarity matrix
2	CMIVGSD (2021)	Graph variational autoencoders	Qian et al. (2021)	circRNA and miRNA similarity matrix
3	GCNCMI (2022)	Graph convolutional neural network	He et al. (2022)	circRNA and miRNA association matrix
4	SGCNCMI (2022)	Graph convolutional neural network	Yu et al. (2022a)	circRNA and miRNA similarity matrix
5	WSCD (2022)	Convolutional neural network and deep neural network	Guo et al. (2022)	circRNA and miRNA similarity matrix
6	IIMCCMA (2022)	Network embedding method	Yao et al. (2022)	circRNA and miRNA similarity matrix
7	KGDMCI (2022)	Deep neural network	Wang et al. (2022b)	K mers, circRNA and miRNA similarity matrix
8	NGCICM (2023)	Graph attention network	Ma et al. (2023)	circRNA and miRNA similarity matrix

circRNA, circular RNA.

Each row of the matrix represents a circRNA, while each column represents an miRNA. A high value in a particular cell suggests a strong probability of interaction between the circRNA and miRNA. For example, the GCNCMI, the interaction graph, contains the higher order interaction information of circRNA and miRNA from which we can mine deep semantic information that carries a collaborative signal (He et al., 2022).

The latest one NGCICM incorporates node2vec and graph attention network, conditional random field layer, and inductive matrix completion, together helping to attain an interaction score. The NGCICM model takes the sequence and expression data of circRNAs and miRNAs as input and uses the above techniques to learn a graph representation of the interactions between circRNAs and miRNAs. The model then makes predictions of new circRNA-miRNA interactions based on this learned representation. According to the authors, NGCICM achieves state-of-the-art performance on several benchmark data sets, with an AUC score of up to 0.982, indicating its high prediction accuracy (Ma et al., 2023). Rest of the two methods are based on the network embedding methods NECMA (Lan et al., 2021) and IIMCCMA (Yao et al., 2022).

Network embedding has gained significant attention in recent years due to its ability to learn low-dimensional representations of complex networks, which can improve efficiency, both these models have a higher accuracy more than 90%.

miRNA and mRNA expression profiling

The difficulty in identifying the tissues in which the level of miRNA is expressed is because great limitations exist in the current expression profile detecting techniques, such as microarray (Bargaje et al., 2010). Because the amount of miRNAs contained in a microarray chip is relatively limited, compared with the total amount of miRNAs in the human genome (Nersisyan et al., 2020), this results in establishing an SVM-based approach called Cepred (Wang et al., 2009a) for predicting the coexpression patterns of the human intronic miRNAs with their host genes. Structural characteristics such as distance from the transcription start position of the host gene to the start point of the host intron and miRNA length of the host intron and miRNA act as the features.

The progress of miRNA-mRNA expression correlation databases such as mirCoX (Giles et al., 2013) and DIANA-miTED (Kavakiotis et al., 2022) lead to the comprehensive analysis of expression profiling. In addition, a new method of predicting the expression patterns with the help of linear regression and relevance vector machine from its promoter sequence was introduced by Oğul and Tuncer (2016). ML-based tool maTE (Yousef et al., 2019) integrates differential gene expression data with miRNA target genes. Integrating miRNA regulation with expression data yields powerful results. It is independent of external labels and training data. Another one, the miRcorrNet (Yousef et al., 2021), combined feature grouping and ranking methods to analyze miRNA and mRNA gene expression profiles.

The applicability of this tool in various biological contexts identifies correlation patterns of significant genes linked with a specific miRNA. The predictive model uses ML algorithms, including random forests, gradient boosting, and SVM. The tools and approaches are listed in Table 7.

Table 7.

MicroRNA-mRNA Expression Profiling

Sl. No	Tool name	Algorithm	Reference
1	miRcorrNet (2021)	Feature grouping and ranking	Yousef et al. (2021)
2	Ogul Hasan et al. (2016)	Relevance vector machine as regression model	Oğul and Tuncer (2016)
3	maTE (2019)	Relevance vector machine	Yousef et al. (2019)
4	Cepred (2009)	Support vector machine	Wang et al. (2009a)

miRNA regulatory module detection

MiRNA regulatory modules are groups of miRNAs whose target genes are coregulated and function together in a biological pathway or process. Identifying miRNA regulatory modules helps decipher the evidence of miRNA and mRNA in gene regulatory networks (Masud Karim et al., 2016). ML methods involve analyzing miRNA and gene expression data and identifying coregulated miRNA gene pairs using miRNA-mRNA target interaction. For example, the clustering miRNA-target interactions is a convolutional autoencoder and affinity propagation-based approach for identifying miRNA regulatory modules by extracting miRNA-target interaction features and their overlapping characteristics from the expression profile data (Yang and Wan, 2020). A similar way of prediction by obtaining a reliable relationship between miRNAs and target genes is DeMosa, and it also has the combined effect of stacked autoencoders and K-means clustering (Yang and Song, 2019).

The deep neural network using the miRModuleNet (Yousef et al., 2022) framework has been applied to identify miRNA-mRNA regulatory modules in various biological processes and diseases. For example, miRModuleNet was used to identify miRNA-mRNA regulatory modules involved in breast cancer progression. The identified miRNA-mRNA regulatory modules were enriched in cell cycle regulation and apoptosis pathways, suggesting their potential roles in breast cancer development. An overview of these algorithms or approaches is summarized in Table 8.

Table 8.

MicroRNA Regulatory Module Detection Tools

Sl. No	Tool name	Algorithm	Reference
1	DeMosa (2019)	Stacked autoencoder	Yang and Song (2019)
2	CMIN (2020)	Convolutional autoencoder	Yang and Wan (2020)
3	miRModuleNet (2022)	Random forest	Yousef et al. (2022)

miRNA disease association prediction

Root causes of human disease development are traced back to miRNA functions in gene expression. Abnormal expression and dysregulations of the miRNA can modulate the metabolic/signaling pathway, which can lead to imbalances in homeostasis. Various bacterial, viral, and protozoan infections can lead to dysregulation of miRNA that significantly affects the signaling pathways influencing the disease outcomes (Agbu and Carthew, 2021; Antil et al., 2022; Ramakrishnan et al., 2023; Ramesh et al., 2023). Understanding disease mechanisms at the miRNA level uncovers the disease biomarkers for diagnosis, medication, prognosis, and prevention (Condrat et al., 2020). For miRNA disease association prediction, the miRNA-disease sequence information and features were majorly taken from searchable databases such as HMDD (Huang et al., 2019b), miR2disease (Jiang et al., 2009), and dbDEMC (Yang et al., 2017).

Data sets having reliable features contribute precision to the output. Prior association of labeled information for investigated miRNAs is required when training an ML model. Most of the approaches such as Jiang et al. (2013), RWRMDA (Chen et al., 2012), EDTMDA (Chen et al., 2019b), and MLMDA (Zheng et al., 2019a) are majorly focused on the functional similarity score information, constructed by integrating miRNA-miRNA interaction data and functional annotation typically depicted as a graph where nodes represent miRNAs and edges represent the functional similarity between miRNAs based on their target genes. Several methods can generate functional similarity scores, including semantic similarity, topology-based similarity, and information content-based similarity. Semantic similarity measures the similarity between miRNA and disease annotations based on their shared ancestor terms in a hierarchical ontology. Topology is based on the number of shared terms and their positions in the ontology hierarchy.

Information content similarity considers the frequency of occurrence of the shared terms in the ontology. Once functional similarity scores are calculated, they can be used to prioritize potential miRNA disease associations. For example, a high functional similarity score between an miRNA and a disease indicates that the miRNA may play a critical role in the pathogenesis of the disease and is likely to be a good candidate for further experimental validation. The heterogeneous graph inference for miRNA-disease association prediction (HGIMDA) (Chen et al., 2016b) is a random walk approach that predicts the association by integrating miRNA functional similarity, disease semantic similarity, Gaussian interaction profile kernel similarity and experimentally verified miRNA-disease associations into a heterogeneous graph. The HGIMDA was applied to human cancers for performance evaluation. An overview of these algorithms or approaches is summarized in Table 9.

Table 9.

MicroRNA-Disease Association Prediction Tools

Sl. No	Tool name	Algorithm	PubMed ID
1	RWRMDA (2012)	Random walk	Chen et al. (2012)
2	RLSMDA (2014)	Regularized least square	Chen and Yan (2014)
3	RBMMMDA (2015)	Restricted Boltzmann machine	Chen et al. (2015)
4	HGIMDA (2016)	Random walk	Chen et al. (2016b)
5	DeepMDA (2017)	Deep ensemble	Fu and Peng (2017)
6	EDTMDA (2019)	Decision tree	Chen et al. (2019b)
7	MLMDA (2019)	Deep autoencoder neural network Random forest classifier	Zheng et al. (2019a)
8	VAEMDA (2019)	Variational autoencoder	Zhang et al. (2019a)
9	PMAMCA (2019)	Matrix factorization	Ha et al. (2019)
10	EKRRMDA (2020)	Ensemble of kernel ridge regression	Peng et al. (2020)
11	Seq-SymRF (2020)	Random forest	Li et al. (2020a)
12	SMALF (2021)	Stacked autoencoder	Liu et al. (2021)
13	NIMGSA (2022)	Autoencoder	Jin et al. (2022)
14	MVMTMDA (2021)	Multiview multitask learning	Huang et al. (2021)
15	NMCMDA (2021)	Neural network	Wang et al. (2021)
16	DANE-MDA (2021)	Deep attributed network embedding	Ji et al. (2021)

Predicting the essentiality of miRNA

Short noncoding RNA molecules known as miRNAs play a crucial role in metabolic processes by posttranscriptionally controlling gene expression. Identifying the important miRNAs in the miRNAome requires predicting their essentiality; however, the bioinformatic techniques available for this purpose are limited. Here, a novel approach called miES (Song et al., 2019) is used to rank the significance of miRNAs. The biological features used in miES include the base pair content in precursor and mature miRNAs, mature and nonmature miRNA lengths, and the secondary structure's thermodynamic properties. The complex and deeper features related to miRNAs improve the essentiality prediction. The PESM (Yan et al., 2020) predicts the miRNA essentiality based on gradient boosting machines and miRNA sequences. Compared with miES, PESM integrates more sequences and structural features of miRNAs such as the base pairing propensity and dinucleotide frequency information of precursor and mature miRNAs.

The list of tools and approaches is summarized in Table 10.

Table 10.

MicroRNA Essentiality Prediction Tools

Sl. No	Tool name	Algorithm	PubMed ID	Salient features
1	miES (2019)	Logistic regression	Song et al. (2019)	Nucleotide content, length of mature and nonmature miRNA, free energy
2	PESM (2020)	Gradient boosting machines	Yan et al. (2020)	Structure, sequence, energy

Conclusions

ML has emerged as a valuable component in the multifaceted analysis of miRNAs. The number of novel miRNA features used for training has exploded with data derived from NGS platforms. Prediction of miRNA function had been most explored computationally based on target prediction. These technological advancements have been instrumental in predicting miRNA targets, analysis of their molecular regulatory networks, investigating concepts of miRNA delivery, and studying epigenetic regulation. These are expected to lead to new insights into miRNA function, as well as the development of miRNA-based therapies and diagnostic tools for a wide range of diseases. ML, most recently deep learning, is widely used, and over time, single methods are combined with integrated platforms to improve the plausibility of the predictions. Although novel miRNA biomarkers are being explored from manually curated information, several pipeline tools have also been developed to analyze miRNA-disease associations.

Data integration is becoming critical as miRNA-associated high-throughput sequencing data grow exponentially.

The most troubling aspects of the current trends were the tool updations or follow-ups and availability of active tools. Most studies are approaches, especially disease association studies, because developing an approach to a tool requires more compelling data, computing efficiency, and accurate maintenance. However, many early tools are still well maintained and frequently updated or integrated with new tools. Predictably, future miRNA ML tools will contain the following characteristics: aim for new miRNA knowledge, analyze high-throughput miRNA technology data, integrate multilevel omics data, and focus on human disease. These collectively highlight the key trends in miRNA ML tool development, researchers can ensure that future miRNA tools are better equipped to meet the needs of the scientific community and address important questions related to miRNA integrative biology.

Key Points and Executive Highlights

The miRNA biogenesis is critical to the posttranscriptional regulation and the emergence of miRNA-based databases diversified the miRNA biology to be visualized at various layers for ML-based applications.

ML approaches have great potential to contribute to the miRNA regulatory network of mRNA, lncRNA, and circRNAs.

Artificial neural network, SVM, and random forest marks a major part of ML in miRNA research.

The lack of availability of reliable tools pertaining to miRNA promoter prediction, target prediction, precursor classification, and disease association is a significant concern in the field.

Development of efficient and well-maintained tools is necessary to effectively extract both sequence and structural features of miRNA from different database resources.

Taken together, this review critically examines and unpacks the cutting-edge synergy of ML approaches and miRNA research so as to develop a dynamic and microlevel understanding of human health and diseases.

Footnotes

Acknowledgment

The authors wish to thank Dr. Ajith V. Pankajam, National Cancer Institute, National Institutes of Health (NIH), USA, for critical comments and feedback and an earlier version of the article.

Author Disclosure Statement

The authors declare they have no conflicting financial interests.

Funding Information

No funding was received for this article.

Abbreviations Used

References

Abeel

, Saeys

, Bonnet

, et al. Generic eukaryotic core promoter prediction using structural features of DNA. Genome Res, 2008; 18(2):310–323; doi: 10.1101/gr.6991408

Adams

, Kasinski

, Slack

. Aberrant regulation and function of microRNAs in cancer. Curr Biol, 2014; 24(16):R762–R776; doi: 10.1016/j.cub.2014.06.043

Agarwal

, Vaz

, Bhattacharya

, et al. Prediction of novel precursor miRNAs using a context-sensitive hidden Markov model (CSHMM). BMC Bioinformatics, 2010; 11(Suppl 1):S29; doi: 10.1186/1471-2105-11-S1-S29

Agbu

, Carthew

. MicroRNA-mediated regulation of glucose and lipid metabolism. Nat Rev Mol Cell Biol, 2021; 22(6):425–438; doi: 10.1038/s41580-021-00354-w

Ahmadi

, Ahmadi

, Azimzadeh-Jamalkandi

, et al. HomoTarget: A new algorithm for prediction of microRNA targets in Homo sapiens. Genomics, 2013; 101(2):94–100; doi: 10.1016/j.ygeno.2012.11.005

Ambros

, Bartel

, et al. A uniform system for microRNA annotation. RNA, 2003; 9(3):277–279; doi: 10.1261/rna.2183803

Antil

, Arefian

, Kandiyil

, et al. The core human microRNAs regulated by Toxoplasma gondii. Microrna, 2022; 11(2):163–174; doi: 10.2174/2211536611666220428130250

Asim

, Ibrahim

, Zehe

, et al. BoT-Net: A lightweight bag of tricks-based neural network for efficient LncRNA-miRNA interaction prediction. Interdiscip Sci, 2022; 14(4):841–862; doi: 10.1007/s12539-022-00535-x

Azari

, Nazari

, Mohit

, et al. Machine learning algorithms reveal potential miRNAs biomarkers in gastric cancer. Sci Rep, 2023; 13(1):6147; doi: 10.1038/s41598-023-32332-x

10.

Backes

, Fehlmann

, Kern

, et al. miRCarta: A central repository for collecting miRNA candidates. Nucleic Acids Res, 2018; 46(D1):D160–D167; doi: 10.1093/nar/gkx851

11.

Bail

, Swerdel

, Liu

, et al. Differential regulation of microRNA stability. RNA, 2010; 16(5):1032–1039; doi: 10.1261/rna.1851510

12.

Bajic

, Tan

, Suzuki

, et al. Promoter prediction analysis on the whole human genome. Nat Biotechnol, 2004; 22(11):1467–1473; doi: 10.1038/nbt1032

13.

Bandyopadhyay

, Mitra

. TargetMiner: MicroRNA target prediction with systematic identification of tissue-specific negative examples. Bioinformatics, 2009; 25(20):2625–2631; doi: 10.1093/bioinformatics/btp503

14.

Bao

, Yang

, Huang

, et al. LncRNADisease 2.0: An updated database of long non-coding RNA-associated diseases. Nucleic Acids Res, 2019; 47(D1):D1034–D1037; doi: 10.1093/nar/gky905

15.

Bargaje

, Hariharan

, Scaria

, et al. Consensus miRNA expression profiles derived from interplatform normalization of microarray data. RNA, 2010; 16(1):16–25; doi: 10.1261/rna.1688110

16.

Bartel

. MicroRNAs: Genomics, biogenesis, mechanism, and function. Cell, 2004; 116(2):281–297; doi: 10.1016/s0092-8674(04)00045-5

17.

Bartel

. MicroRNAs: Target recognition and regulatory functions. Cell, 2009; 136(2):215–233; doi: 10.1016/j.cell.2009.01.002

18.

Batuwita

, Palade

. microPred: Effective classification of pre-miRNAs for human miRNA gene prediction. Bioinformatics, 2009; 25(8):989–995; doi: 10.1093/bioinformatics/btp107

19.

Bentwich

, Avniel

, Karov

, et al. Identification of hundreds of conserved and nonconserved human microRNAs. Nat Genet, 2005; 37(7):766–770; doi: 10.1038/ng1590

20.

Berezikov

, Guryev

, van de Belt

, et al. Phylogenetic shadowing and computational identification of human microRNA genes. Cell, 2005; 120(1):21–24; doi: 10.1016/j.cell.2004.12.031

21.

Bhattacharyya

, Das

, Bandyopadhyay

. miRT: A database of validated transcription start sites of human microRNAs. Genomics Proteomics Bioinformatics, 2012a;10(5):310–316; doi: 10.1016/j.gpb.2012.08.005

22.

Bhattacharyya

, Feuerbach

, Bhadra

, et al. MicroRNA transcription start site prediction with multi-objective feature selection. Stat Appl Genet Mol Biol, 2012b;11(1):Article 6; doi: 10.2202/1544-6115.1743

23.

Bracken

, Gregory

, Kolesnikoff

, et al. A double-negative feedback loop between ZEB1-SIP1 and the microRNA-200 family regulates epithelial-mesenchymal transition. Cancer Res, 2008; 68(19):7846–7854; doi: 10.1158/0008-5472.CAN-08-1942

24.

Bueno

, Perez de Castro

, Gomez de Cedron

, et al. Genetic and epigenetic silencing of microRNA-203 enhances ABL1 and BCR-ABL1 oncogene expression. Cancer Cell, 2008; 13(6):496–506; doi: 10.1016/j.ccr.2008.04.018

25.

Cai

, Gao

, Ren

, et al. MILNP: Plant lncRNA-miRNA interaction prediction based on improved linear neighborhood similarity and label propagation. Front Plant Sci, 2022; 13:861886; doi: 10.3389/fpls.2022.861886

26.

Cai

, Hagedorn

, Cullen

. Human microRNAs are processed from capped, polyadenylated transcripts that can also function as mRNAs. RNA, 2004; 10(12):1957–1966; doi: 10.1261/rna.7135204

27.

Cesana

, Cacchiarelli

, Legnini

, et al. A long noncoding RNA controls muscle differentiation by functioning as a competing endogenous RNA. Cell, 2011; 147(2):358–369; doi: 10.1016/j.cell.2011.09.028

28.

Cha

, Zheng

, Talukder

, et al. A two-stream convolutional neural network for microRNA transcription start site feature integration and identification. Sci Rep, 2021; 11(1):5625; doi: 10.1038/s41598-021-85173-x

29.

Chandra

, Girijadevi

, Nair

, et al. MTar: A computational microRNA target prediction architecture for human transcriptome. BMC Bioinformatics, 2010; 11(Suppl 1):S2; doi: 10.1186/1471-2105-11-S1-S2

30.

Chang

, Wang

, Chen

. Using a kernel density estimation based classifier to predict species-specific microRNA precursors. BMC Bioinformatics, 2008a;9(Suppl 12):S2; doi: 10.1186/1471-2105-9-S12-S2

31.

Chang

, Yu

, Lee

, et al. Widespread microRNA repression by Myc contributes to tumorigenesis. Nat Genet, 2008b;40(1):43–50; doi: 10.1038/ng.2007.30

32.

Chen

, Heikkinen

, Wang

, et al. Trends in the development of miRNA bioinformatics tools. Brief Bioinform, 2019a;20(5):1836–1852; doi: 10.1093/bib/bby054

33.

Chen

, Han

, Zhou

, et al. circRNADb: A comprehensive database for human circular RNAs with protein-coding annotations. Sci Rep, 2016a;6:34985; doi: 10.1038/srep34985

34.

Chen

, Liu

, Yan

. RWRMDA: Predicting novel human microRNA-disease associations. Mol Biosyst, 2012; 8(10):2792–2798; doi: 10.1039/c2mb25180a

35.

Chen

, Yan

, Zhang

, et al. RBMMMDA: Predicting multiple types of disease-microRNA associations. Sci Rep, 2015; 5:13877; doi: 10.1038/srep13877

36.

Chen

, Yan

, Zhang

, et al. HGIMDA: Heterogeneous graph inference for miRNA-disease association prediction. Oncotarget, 2016b;7(40):65257–65269; doi: 10.18632/oncotarget.11251

37.

Chen

, Yan

. Semi-supervised learning for potential human microRNA-disease associations inference. Sci Rep, 2014; 4:5501; doi: 10.1038/srep05501

38.

Chen

, Zhu

, Yin

. Ensemble of decision tree reveals potential miRNA-disease associations. PLoS Comput Biol, 2019b;15(7):e1007209; doi: 10.1371/journal.pcbi.1007209

39.

Chen

, Wang

. miRDB: An online database for prediction of functional microRNA targets. Nucleic Acids Res, 2020; 48(D1):D127–D131; doi: 10.1093/nar/gkz757

40.

Chien

, Sun

, Chang

, et al. Identifying transcriptional start sites of human microRNAs based on high-throughput sequencing data. Nucleic Acids Res, 2011; 39(21):9345–9356; doi: 10.1093/nar/gkr604

41.

Chipman

, Pasquinelli

. miRNA targeting: Growing beyond the seed. Trends Genet, 2019; 35(3):215–222; doi: 10.1016/j.tig.2018.12.005

42.

Chong

, Zhang

, Cheloufi

, et al. Canonical and alternate functions of the microRNA biogenesis machinery. Genes Dev, 2010; 24(17):1951–1960; doi: 10.1101/gad.1953310

43.

Chu

, Qu

, Zhong

, et al. Genomic maps of long noncoding RNA occupancy reveal principles of RNA-chromatin interactions. Mol Cell, 2011; 44(4):667–678; doi: 10.1016/j.molcel.2011.08.027

44.

Condrat

, Thompson

, Barbu

, et al. miRNAs as biomarkers in disease: Latest findings regarding their role in diagnosis and prognosis. Cells, 2020; 9(2):276; doi: 10.3390/cells9020276

45.

Cui

, Zhai

, Ma

. miRLocator: Machine learning-based prediction of mature microRNAs within plant pre-miRNA sequences. PLoS One, 2015; 10(11):e0142753; doi: 10.1371/journal.pone.0142753

46.

Cullen

. Transcription and processing of human microRNA precursors. Mol Cell, 2004; 16(6):861–865; doi: 10.1016/j.molcel.2004.12.002

47.

Davis-Dusenbery

, Hata

. Mechanisms of control of microRNA biogenesis. J Biochem, 2010; 148(4):381–392; doi: 10.1093/jb/mvq096

48.

Ding

, Li

, Hu

. TarPmiR: A new approach for microRNA target site prediction. Bioinformatics, 2016; 32(18):2768–2775; doi: 10.1093/bioinformatics/btw318

49.

Ding

, Zhou

, Guan

. MiRenSVM: Towards better prediction of microRNA precursors using an ensemble SVM classifier with multi-loop features. BMC Bioinformatics, 2010; 11(Suppl 11):S11; doi: 10.1186/1471-2105-11-S11-S11

50.

Dong

, Ma

, Li

, et al. CIRCpedia v2: An updated database for comprehensive circular RNA annotation and expression comparison. Genomics Proteomics Bioinformatics, 2018; 16(4):226–233; doi: 10.1016/j.gpb.2018.08.001

51.

Enright

, John

, Gaul

, et al. MicroRNA targets in Drosophila. Genome Biol, 2003; 5(1):R1; doi: 10.1186/gb-2003-5-1-r1

52.

Faller

, Matsunaga

, Yin

, et al. Heme is involved in microRNA processing. Nat Struct Mol Biol, 2007; 14(1):23–29; doi: 10.1038/nsmb1182

53.

Fiannaca

, Rosa

, Paglia

, et al. MiRNATIP: A SOM-based miRNA-target interactions predictor. BMC Bioinformatics, 2016; 17(Suppl 11):321; doi: 10.1186/s12859-016-1171-x

54.

, Peng

. A deep ensemble model to predict miRNA-disease association. Sci Rep, 2017; 7(1):14482; doi: 10.1038/s41598-017-15235-6

55.

Gao

, Middleton

, Rasko

, et al. miREval 2.0: A web tool for simple microRNA prediction in genome sequences. Bioinformatics, 2013; 29(24):3225–3226; doi: 10.1093/bioinformatics/btt545

56.

Georgakilas

, Vlachos

, Paraskevopoulou

, et al. microTSS: Accurate microRNA transcription start site identification reveals a significant number of divergent pri-miRNAs. Nat Commun, 2014; 5:5700; doi: 10.1038/ncomms6700

57.

Gerlach

, Kriventseva

, Rahman

, et al. miROrtho: Computational survey of microRNA genes. Nucleic Acids Res, 2009; 37(Database Issue):D111–D117; doi: 10.1093/nar/gkn707

58.

Giles

, Girija-Devi

, Dozmorov

, et al. mirCoX: A database of miRNA-mRNA expression correlations derived from RNA-seq meta-analysis. BMC Bioinformatics, 2013; 14(Suppl 14):S17; doi: 10.1186/1471-2105-14-S14-S17

59.

Gkirtzou

, Tsamardinos

, Tsakalides

, et al. MatureBayes: A probabilistic algorithm for identifying the mature miRNA within novel precursors. PLoS One, 2010; 5(8):e11843; doi: 10.1371/journal.pone.0011843

60.

Glazar

, Papavasileiou

, Rajewsky

. circBase: A database for circular RNAs. RNA, 2014; 20(11):1666–1670; doi: 10.1261/rna.043687.113

61.

Gong

, Liu

, Zhang

, et al. lncRNASNP: A database of SNPs in lncRNAs and their potential functions in human and mouse. Nucleic Acids Res, 2015; 43(Database Issue):D181–D186; doi: 10.1093/nar/gku1000

62.

Gregory

, Yan

, Amuthan

, et al. The microprocessor complex mediates the genesis of microRNAs. Nature, 2004; 432(7014):235–240; doi: 10.1038/nature03120

63.

Griffiths-Jones

, Grocock

, van Dongen

, et al. miRBase: MicroRNA sequences, targets and gene nomenclature. Nucleic Acids Res, 2006; 34(Database Issue):D140–D144; doi: 10.1093/nar/gkj112

64.

, Zhao

, Barbazuk

, et al. miTAR: A hybrid deep learning-based approach for predicting miRNA targets. BMC Bioinformatics, 2021; 22(1):96; doi: 10.1186/s12859-021-04026-6

65.

Gudys

, Szczesniak

, Sikora

, et al. HuntMi: An efficient and taxon-specific approach in pre-miRNA identification. BMC Bioinformatics, 2013; 14:83; doi: 10.1186/1471-2105-14-83

66.

Guerra-Assuncao

, Enright

. MapMi: Automated mapping of microRNA loci. BMC Bioinformatics, 2010; 11:133; doi: 10.1186/1471-2105-11-133

67.

Guo

, You

, Wang

, et al. A novel circRNA-miRNA association prediction model based on structural deep neural network embedding. Brief Bioinform, 2022; 23(5):bbac391; doi: 10.1093/bib/bbac391

68.

, Park

. PMAMCA: Prediction of microRNA-disease association utilizing a matrix completion approach. BMC Syst Biol, 2019; 13(1):33; doi: 10.1186/s12918-019-0700-4

69.

, Kim

. Regulation of microRNA biogenesis. Nat Rev Mol Cell Biol, 2014; 15(8):509–524; doi: 10.1038/nrm3838

70.

Han

, Lee

, Yeom

, et al. Molecular basis for the recognition of primary microRNAs by the Drosha-DGCR8 complex. Cell, 2006; 125(5):887–901; doi: 10.1016/j.cell.2006.03.043

71.

Hansen

, Jensen

, Clausen

, et al. Natural RNA circles function as efficient microRNA sponges. Nature, 2013; 495(7441):384–388; doi: 10.1038/nature11993

72.

, Li

, Zhang

, et al. MiRmat: Mature microRNA sequence prediction. PLoS One, 2012; 7(12):e51673; doi: 10.1371/journal.pone.0051673

73.

, Xiao

, Chen

, et al. GCNCMI: A graph convolutional neural network approach for predicting circRNA-miRNA interactions. Front Genet, 2022; 13:959701; doi: 10.3389/fgene.2022.959701

74.

Heikkinen

, Kolehmainen

, Wong

. Prediction of microRNA targets in Caenorhabditis elegans using a self-organizing map. Bioinformatics, 2011; 27(9):1247–1254; doi: 10.1093/bioinformatics/btr144

75.

Hertel

, Langenberger

, Stadler

. Computational prediction of microRNA genes. Methods Mol Biol, 2014; 1097:437–456; doi: 10.1007/978-1-62703-709-9_20

76.

Hertel

, Stadler

. Hairpins in a Haystack: Recognizing microRNA precursors in comparative genomics data. Bioinformatics, 2006; 22(14):e197–e202; doi: 10.1093/bioinformatics/btl257

77.

Hibio

, Hino

, Shimizu

, et al. Stability of miRNA 5'terminal and seed regions is correlated with experimentally observed miRNA-mediated silencing efficacy. Sci Rep, 2012; 2:996; doi: 10.1038/srep00996

78.

Hollbacher

, Balazs

, Heinig

, et al. Seq-ing answers: Current data integration approaches to uncover mechanisms of transcriptional regulation. Comput Struct Biotechnol J, 2020; 18:1330–1341; doi: 10.1016/j.csbj.2020.05.018

79.

Hsu

, Coca-Prados

. Electron microscopic evidence for the circular form of RNA in the cytoplasm of eukaryotic cells. Nature, 1979; 280(5720):339–340; doi: 10.1038/280339a0

80.

Hsu

, Lin

, Wu

, et al. miRTarBase: A database curates experimentally validated microRNA-target interactions. Nucleic Acids Res, 2011; 39(Database Issue):D163–D169; doi: 10.1093/nar/gkq1107

81.

, Huang

, Chan

KCC

, et al. Learning multimodal networks from heterogeneous data for prediction of lncRNA-miRNA interactions. IEEE/ACM Trans Comput Biol Bioinform, 2020; 17(5):1516–1524; doi: 10.1109/TCBB.2019.2957094

82.

Huang

, Fan

, Rothschild

, et al. MiRFinder: An improved approach and software implementation for genome-wide fast microRNA precursor scans. BMC Bioinformatics, 2007; 8:341; doi: 10.1186/1471-2105-8-341

83.

Huang

, Chan

KCC

, You

. Constructing prediction models from expression profiles for large scale lncRNA-miRNA interaction profiling. Bioinformatics, 2018a;34(5):812–819; doi: 10.1093/bioinformatics/btx672

84.

Huang

, Chan

KCC

, You

, et al. Predicting microRNA-disease associations from lncRNA-microRNA interactions via multiview multitask learning. Brief Bioinform, 2021; 22(3):bbaa133; doi: 10.1093/bib/bbaa133

85.

Huang

, Huang

, You

, et al. Predicting lncRNA-miRNA interaction via graph convolution auto-encoder. Front Genet, 2019a;10:758; doi: 10.3389/fgene.2019.00758

86.

Huang

, Shi

, Gao

, et al. HMDD v3.0: A database for experimentally supported human microRNA-disease associations. Nucleic Acids Res, 2019b;47(D1):D1013–D1017; doi: 10.1093/nar/gky1010

87.

Huang

, Huang

, You

, et al. Novel link prediction for large-scale miRNA-lncRNA interaction network in a bipartite graph. BMC Med Genomics, 2018b;11(Suppl 6):113; doi: 10.1186/s12920-018-0429-8

88.

Hwang

, Mendell

. MicroRNAs in cell proliferation, cell death, and tumorigenesis. Br J Cancer, 2006; 94(6):776–780; doi: 10.1038/sj.bjc.6603023

89.

Isoda

, Moore

, He

, et al. Non-coding transcription instructs chromatin folding and compartmentalization to dictate enhancer-promoter communication and T cell fate. Cell, 2017; 171(1):103–119 e118; doi: 10.1016/j.cell.2017.09.001

90.

, You

, Wang

, et al. DANE-MDA: Predicting microRNA-disease associations via deep attributed network embedding. iScience, 2021; 24(6):102455; doi: 10.1016/j.isci.2021.102455

91.

Jiang

, Wu

, Wang

, et al. MiPred: Classification of real and pseudo microRNA precursors using random forest prediction model with combined features. Nucleic Acids Res, 2007; 35(Web Server Issue):W339–W344; doi: 10.1093/nar/gkm368

92.

Jiang

, Wang

, Jin

, et al. Predicting human microRNA-disease associations based on support vector machine. Int J Data Min Bioinform, 2013; 8(3):282–293; doi: 10.1504/ijdmb.2013.056078

93.

Jiang

, Wang

, Hao

, et al. miR2Disease: A manually curated database for microRNA deregulation in human disease. Nucleic Acids Res, 2009; 37(Database Issue):D98–D104; doi: 10.1093/nar/gkn714

94.

Jin

, Shi

, Lin

, et al. Predicting miRNA-disease association based on neural inductive matrix completion with graph autoencoders and self-attention mechanism. Biomolecules, 2022; 12(1):64; doi: 10.3390/biom12010064

95.

Jones-Rhoades

. Prediction of plant miRNA genes. Methods Mol Biol, 2010; 592:19–30; doi: 10.1007/978-1-60327-005-2_2

96.

Kallen

, Zhou

, Xu

, et al. The imprinted H19 lncRNA antagonizes let-7 microRNAs. Mol Cell, 2013; 52(1):101–112; doi: 10.1016/j.molcel.2013.08.027

97.

Kang

, Meng

, Cui

, et al. PmliPred: A method based on hybrid model and fuzzy decision for plant miRNA-lncRNA interaction prediction. Bioinformatics, 2020; 36(10):2986–2992; doi: 10.1093/bioinformatics/btaa074

98.

Kang

, Meng

, Shi

, et al. Ensemble deep learning based on multi-level information enhancement and greedy fuzzy decision for plant miRNA-lncRNA interaction prediction. Interdiscip Sci, 2021; 13(4):603–614; doi: 10.1007/s12539-021-00434-7

99.

Karagkouni

, Paraskevopoulou

, Tastsoglou

, et al. DIANA-LncBase v3: Indexing experimentally supported miRNA targets on non-coding transcripts. Nucleic Acids Res, 2020; 48(D1):D101–D110; doi: 10.1093/nar/gkz1036

100.

Karathanasis

, Tsamardinos

, Poirazi

. MiRduplexSVM: A high-performing miRNA-duplex prediction and evaluation methodology. PLoS One, 2015; 10(5):e0126151; doi: 10.1371/journal.pone.0126151

101.

Kavakiotis

, Alexiou

, Tastsoglou

, et al. DIANA-miTED: A microRNA tissue expression database. Nucleic Acids Res, 2022; 50(D1):D1055–D1061; doi: 10.1093/nar/gkab733

102.

Kawamata

, Seitz

, Tomari

. Structural determinants of miRNAs for RISC loading and slicer-independent unwinding. Nat Struct Mol Biol, 2009; 16(9):953–960; doi: 10.1038/nsmb.1630

103.

Kim

S-K

, Nam

J-W

, Lee

W-J

, et al. A Kernel method for microRNA target prediction using sensible data and position-based features. In: 2005 IEEE Symposium on Computational Intelligence in Bioinformatics and Computational Biology. IEEE: 2005; pp. 1–7.

104.

Kim

, Nam

, Rhee

, et al. miTarget: MicroRNA target gene prediction using a support vector machine. BMC Bioinformatics, 2006; 7:411; doi: 10.1186/1471-2105-7-411

105.

Kim

. MicroRNA biogenesis: Coordinated cropping and dicing. Nat Rev Mol Cell Biol, 2005; 6(5):376–385; doi: 10.1038/nrm1644

106.

Kim

, Han

, Siomi

. Biogenesis of small RNAs in animals. Nat Rev Mol Cell Biol, 2009; 10(2):126–139; doi: 10.1038/nrm2632

107.

Kozomara

, Birgaoanu

, Griffiths-Jones

. miRBase: From microRNA sequences to function. Nucleic Acids Res, 2019; 47(D1):D155–D162; doi: 10.1093/nar/gky1141

108.

Krek

, Grun

, Poy

, et al. Combinatorial microRNA target predictions. Nat Genet, 2005; 37(5):495–500; doi: 10.1038/ng1536

109.

Kristensen

, Andersen

, Stagsted

LVW

, et al. The biogenesis, biology and characterization of circular RNAs. Nat Rev Genet, 2019; 20(11):675–691; doi: 10.1038/s41576-019-0158-7

110.

Krol

, Loedige

, Filipowicz

. The widespread regulation of microRNA biogenesis, function and decay. Nat Rev Genet, 2010; 11(9):597–610; doi: 10.1038/nrg2843

111.

Lagos-Quintana

, Rauhut

, Lendeckel

, et al. Identification of novel genes coding for small expressed RNAs. Science, 2001; 294(5543):853–858; doi: 10.1126/science.1064921

112.

Lan

, Zhu

, Chen

, et al. CircR2Cancer: A manually curated database of associations between circRNAs and cancers. Database (Oxford), 2020; 2020:baaa085; doi: 10.1093/database/baaa085

113.

Lan

, Zhu

, Chen

, et al. Prediction of circRNA-miRNA associations based on network embedding. Complexity, 2021; 2021:1–10.

114.

Lau

, Lim

, Weinstein

, et al. An abundant class of tiny RNAs with probable regulatory roles in Caenorhabditis elegans. Science, 2001; 294(5543):858–862; doi: 10.1126/science.1065062

115.

Lauria

, Curcio

, Tucci

. A machine learning approach for highlighting microRNAs as biomarkers linked to amyotrophic lateral sclerosis diagnosis and progression. Biomolecules, 2023; 14(1):47; doi: 10.3390/biom14010047

116.

Leclercq

, Diallo

, Blanchette

. Computational prediction of the localization of microRNAs within their pre-miRNA. Nucleic Acids Res, 2013; 41(15):7200–7211; doi: 10.1093/nar/gkt466

117.

Lee

, Shin

. MicroRNA-target interactions: New insights from genome-wide approaches. Ann N Y Acad Sci, 2012; 1271(1):118–128; doi: 10.1111/j.1749-6632.2012.06745.x

118.

Lee

, Ahn

, Han

, et al. The nuclear RNase III Drosha initiates microRNA processing. Nature, 2003; 425(6956):415–419; doi: 10.1038/nature01957

119.

Lee

, Kim

, Han

, et al. MicroRNA genes are transcribed by RNA polymerase II. EMBO J, 2004; 23(20):4051–4060; doi: 10.1038/sj.emboj.7600385

120.

Lewis

, Burge

, Bartel

. Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell, 2005; 120(1):15–20; doi: 10.1016/j.cell.2004.12.035

121.

Lewis

, Shih

, Jones-Rhoades

, et al. Prediction of mammalian microRNA targets. Cell, 2003; 115(7):787–798; doi: 10.1016/s0092-8674(03)01018-3

122.

, Chen

, Huang

, et al. Seq-SymRF: A random forest model predicts potential miRNA-disease associations based on information of sequences and clinical symptoms. Sci Rep, 2020a;10(1):17901; doi: 10.1038/s41598-020-75005-9

123.

, Wang

, et al. MatPred: Computational identification of mature microRNAs within novel pre-microRNAs. Biomed Res Int, 2015; 2015:546763; doi: 10.1155/2015/546763

124.

, Nguyen

, et al. Mismatched and wobble base pairs govern primary microRNA processing by human Microprocessor. Nat Commun, 2020b;11(1):1926; doi: 10.1038/s41467-020-15674-2

125.

Liu

, Fang

, Chen

, et al. miRNA-dis: MicroRNA precursor identification based on distance structure status pairs. Mol Biosyst, 2015; 11(4):1194–1204; doi: 10.1039/c5mb00050e

126.

Liu

, Li

, Cairns

. Identifying miRNAs, targets and functions. Brief Bioinform, 2014; 15(1):1–19; doi: 10.1093/bib/bbs075

127.

Liu

, Huang

, Nie

, et al. SMALF: MiRNA-disease associations prediction based on stacked autoencoder and XGBoost. BMC Bioinformatics, 2021; 22(1):219; doi: 10.1186/s12859-021-04135-2

128.

Liu

, Yue

, Zhang

, et al. A Machine Learning Approach for miRNA Target Prediction. IEEE Int Workshop Genomic Signal Process Stat, 2008; 2008:1–3; doi: 10.1109/GENSIPS.2008.4555655

129.

Liu

, Ye

, Jiang

, et al. C3PO, an endoribonuclease that promotes RNAi by facilitating RISC activation. Science, 2009; 325(5941):750–753; doi: 10.1126/science.1176325

130.

Lokuge

, Jayasundara

, Ihalagedara

, et al. miRNAFinder: A comprehensive web resource for plant Pre-microRNA classification. Biosystems, 2022; 215–216:104662; doi: 10.1016/j.biosystems.2022.104662

131.

Lujambio

, Calin

, Villanueva

, et al. A microRNA DNA methylation signature for human cancer metastasis. Proc Natl Acad Sci U S A, 2008; 105(36):13556–13561; doi: 10.1073/pnas.0803055105

132.

, Yuan

, Meister

, et al. Structural basis for 5′-end-specific recognition of guide RNA by the A. fulgidus Piwi protein. Nature, 2005; 434(7033):666–670; doi: 10.1038/nature03514

133.

, Kuang

, Deng

. NGCICM: A novel deep learning-based method for predicting circRNA-miRNA interactions. IEEE/ACM Trans Comput Biol Bioinform, 2023; 20(5):3080–3092; doi: 10.1109/TCBB.2023.3248787

134.

Marceca

, Distefano

, Tomasello

, et al. MiREDiBase, a manually curated database of validated and putative editing events in microRNAs. Sci Data, 2021; 8(1):199; doi: 10.1038/s41597-021-00979-8

135.

Masud Karim

, Liu

, Le

, et al. Identification of miRNA-mRNA regulatory modules by exploring collective group relationships. BMC Genomics, 2016; 17(Suppl 1):7; doi: 10.1186/s12864-015-2300-z

136.

Mathews

, Sabina

, Zuker

, et al. Expanded sequence dependence of thermodynamic parameters improves prediction of RNA secondary structure. J Mol Biol, 1999; 288(5):911–940; doi: 10.1006/jmbi.1999.2700

137.

Megraw

, Sethupathy

, Corda

, et al. miRGen: A database for the study of animal microRNA genomic organization and function. Nucleic Acids Res, 2007; 35(Database Issue):D149–D155; doi: 10.1093/nar/gkl904

138.

Megret

, Mendoza

, Arrieta Lobo

, et al. Precision machine learning to understand micro-RNA regulation in neurodegenerative diseases. Front Mol Neurosci, 2022; 15:914830; doi: 10.3389/fnmol.2022.914830

139.

Memczak

, Jens

, Elefsinioti

, et al. Circular RNAs are a large class of animal RNAs with regulatory potency. Nature, 2013; 495(7441):333–338; doi: 10.1038/nature11928

140.

Mendoza

, da Fonseca

, Loss-Morais

, et al. RFMirTarget: Predicting human microRNA target genes with a random forest classifier. PLoS One, 2013; 8(7):e70153; doi: 10.1371/journal.pone.0070153

141.

Menor

, Ching

, Zhu

, et al. mirMark: A site-level and UTR-level classifier for miRNA target prediction. Genome Biol, 2014; 15(10):500; doi: 10.1186/s13059-014-0500-5

142.

Nersisyan

, Shkurnikov

, Poloznikov

, et al. A post-processing algorithm for miRNA microarray data. Int J Mol Sci, 2020; 21(4):1228; doi: 10.3390/ijms21041228

143.

Noble

. What is a support vector machine?. Nat Biotechnol, 2006; 24(12):1565–1567; doi: 10.1038/nbt1206-1565

144.

Oğul

, Tuncer

. MicroRNA expression prediction: Regression from regulatory elements. Biocybern Biomed Eng, 2016; 36(1):89–94.

145.

Ohler

, Niemann

. Identification and analysis of eukaryotic promoters: Recent computational approaches. Trends Genet, 2001; 17(2):56–60; doi: 10.1016/s0168-9525(00)02174-0

146.

Okamura

, Ishizuka

, Siomi

, et al. Distinct roles for Argonaute proteins in small RNA-directed RNA cleavage pathways. Genes Dev, 2004; 18(14):1655–1666; doi: 10.1101/gad.1210204

147.

Ozsolak

, Poling

, Wang

, et al. Chromatin structure analyses identify miRNA promoters. Genes Dev, 2008; 22(22):3172–3183; doi: 10.1101/gad.1706508

148.

Paraskevopoulou

, Hatzigeorgiou

. Analyzing miRNA-lncRNA interactions. Methods Mol Biol, 2016; 1402:271–286; doi: 10.1007/978-1-4939-3378-5_21

149.

Parveen

, Mustafa

, Yadav

, et al. Applications of machine learning in miRNA discovery and target prediction. Curr Genomics, 2019; 20(8):537–544; doi: 10.2174/1389202921666200106111813

150.

Pawelka

, Laczmanska

, Karpinski

, et al. Machine-learning-based analysis identifies miRNA expression profile for diagnosis and prediction of colorectal cancer: A preliminary study. Cancer Genomics Proteomics, 2022; 19(4):503–511; doi: 10.21873/cgp.20336

151.

Peng

, Zhou

, Chen

, et al. A computational study of potential miRNA-disease association inference based on ensemble learning and kernel ridge regression. Front Bioeng Biotechnol, 2020; 8:40; doi: 10.3389/fbioe.2020.00040

152.

Peng

, Croce

. The role of microRNAs in human cancer. Signal Transduct Target Ther, 2016; 1:15004; doi: 10.1038/sigtrans.2015.4

153.

Peterson

, Thompson

, Ufkin

, et al. Common features of microRNA target prediction tools. Front Genet, 2014; 5:23; doi: 10.3389/fgene.2014.00023

154.

Pian

, Zhang

, Tu

, et al. LncCeRBase: A database of experimentally validated human competing endogenous long non-coding RNAs. Database (Oxford), 2018; 2018:bay061; doi: 10.1093/database/bay061

155.

Piletic

, Kunej

. MicroRNA-target interactions reloaded: Identification of potentially functional sequence variants within validated microRNA-target interactions. OMICS, 2018; 22(11):700–708; doi: 10.1089/omi.2018.0159

156.

Piriyapongsa

, Bootchai

, Ngamphiw

, et al. microPIR: An integrated database of microRNA target sites within human promoter sequences. PLoS One, 2012; 7(3):e33888; doi: 10.1371/journal.pone.0033888

157.

Pla

, Zhong

, Rayner

. miRAW: A deep learning-based approach to predict microRNA targets by analyzing whole microRNA transcripts. PLoS Comput Biol, 2018; 14(7):e1006185; doi: 10.1371/journal.pcbi.1006185

158.

Qian

, Zheng

, Zhang

, et al. CMIVGSD: CircRNA-miRNA Interaction Prediction Based on Variational Graph Auto-Encoder and Singular Value Decomposition. In: 2021 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). IEEE: 2021; pp. 205–210.

159.

Raad

, Bugnon

, Milone

, et al. miRe2e: A full end-to-end deep model based on transformers for prediction of pre-miRNAs. Bioinformatics, 2022; 38(5):1191–1197; doi: 10.1093/bioinformatics/btab823

160.

Rahman

, Islam

, et al. MiRANN: A reliable approach for improved classification of precursor microRNA using Artificial Neural Network model. Genomics, 2012; 99(4):189–194; doi: 10.1016/j.ygeno.2012.02.001

161.

Ramakrishnan

, Babu

, Shaji

, et al. Hepatitis B virus modulated transcriptional regulatory map of hepatic cellular microRNAs. OMICS, 2023; 27(12):581–597; doi: 10.1089/omi.2023.0171

162.

Ramesh

, Babu

, Ammankallu

, et al. Helicobacter pylori regulated microRNA map of human gastric cells. Helicobacter, 2023; 28(1):e12941; doi: 10.1111/hel.12941

163.

Rehmsmeier

, Steffen

, Hochsmann

, et al. Fast and effective prediction of microRNA/target duplexes. RNA, 2004; 10(10):1507–1517; doi: 10.1261/rna.5248604

164.

Reyes-Herrera

, Ficarra

. One decade of development and evolution of microRNA target prediction algorithms. Genomics Proteomics Bioinformatics, 2012; 10(5):254–263; doi: 10.1016/j.gpb.2012.10.001

165.

Rojas

AML

, Drusin

, Chorostecki

, et al. Identification of key sequence features required for microRNA biogenesis in plants. Nat Commun, 2020; 11(1):5320; doi: 10.1038/s41467-020-19129-6

166.

Rophina

, Sharma

, Poojary

, et al. Circad: A comprehensive manually curated resource of circular RNA associated with diseases. Database (Oxford), 2020; 2020:baaa019; doi: 10.1093/database/baaa019

167.

Sacar

, Hamzeiy

, Allmer

. Can MiRBase provide positive data for machine learning for the detection of MiRNA hairpins?. J Integr Bioinform, 2013; 10(2):215; doi: 10.2390/biecoll-jib-2013-215

168.

Saetrom

, Snove

Jr. , Saetrom

. Weighted sequence motifs as an improved seeding step in microRNA target prediction algorithms. RNA, 2005; 11(7):995–1003; doi: 10.1261/rna.7290705

169.

Sanger

, Klotz

, Riesner

, et al. Viroids are single-stranded covalently closed circular RNA molecules existing as highly base-paired rod-like structures. Proc Natl Acad Sci U S A, 1976; 73(11):3852–3856; doi: 10.1073/pnas.73.11.3852

170.

Schirle

, MacRae

. The crystal structure of human Argonaute2. Science, 2012; 336(6084):1037–1040; doi: 10.1126/science.1221551

171.

Scott

, Mattie

, Berger

, et al. Rapid alteration of microRNA levels by histone deacetylase inhibition. Cancer Res, 2006; 66(3):1277–1281; doi: 10.1158/0008-5472.CAN-05-3632

172.

Seunghyun

, Min

, Choi

, et al. deepMiRGene: Deep Neural Network based Precursor microRNA Prediction. arXiv, 2016; 2016:arXiv:1605.00017.

173.

Sewer

, Paul

, Landgraf

, et al. Identification of clustered microRNAs using an ab initio prediction method. BMC Bioinformatics, 2005; 6:267; doi: 10.1186/1471-2105-6-267

174.

Shuang

, Maozu

, Chunyu

, et al. MiRTDL: A deep learning approach for miRNA target prediction. IEEE/ACM Trans Comput Biol Bioinform, 2016; 13(6):1161–1169; doi: 10.1109/TCBB.2015.2510002

175.

Siomi

, Siomi

. Posttranscriptional regulation of microRNA biogenesis in animals. Mol Cell, 2010; 38(3):323–332; doi: 10.1016/j.molcel.2010.03.013

176.

Song

, Cui

, Gao

, et al. miES: Predicting the essentiality of miRNAs with machine learning and sequence features. Bioinformatics, 2019; 35(6):1053–1054; doi: 10.1093/bioinformatics/bty738

177.

Song

, Tian

, Yu

, et al. MD-MLI: Prediction of miRNA-lncRNA interaction by using multiple features and hierarchical deep learning. IEEE/ACM Trans Comput Biol Bioinform, 2022; 19(3):1724–1733; doi: 10.1109/TCBB.2020.3034922

178.

Starega-Roslan

, Koscianska

, Kozlowski

, et al. The role of the precursor structure in the biogenesis of microRNA. Cell Mol Life Sci, 2011; 68(17):2859–2871; doi: 10.1007/s00018-011-0726-2

179.

Statello

, Guo

, Chen

, et al. Gene regulation by long non-coding RNAs and its biological functions. Nat Rev Mol Cell Biol, 2021; 22(2):96–118; doi: 10.1038/s41580-020-00315-9

180.

Sturm

, Hackenberg

, Langenberger

, et al. TargetSpy: A supervised machine learning approach for microRNA target prediction. BMC Bioinformatics, 2010; 11:292; doi: 10.1186/1471-2105-11-292

181.

Suzuki

, Arase

, Matsuyama

, et al. MCPIP1 ribonuclease antagonizes dicer and terminates microRNA biogenesis through precursor microRNA degradation. Mol Cell, 2011; 44(3):424–436; doi: 10.1016/j.molcel.2011.09.012

182.

Talukder

, Zhang

, Li

, et al. A deep learning method for miRNA/isomiR target detection. Sci Rep, 2022; 12(1):10618; doi: 10.1038/s41598-022-14890-8

183.

Tarasov

, Jung

, Verdoodt

, et al. Differential regulation of microRNAs by p53 revealed by massively parallel sequencing: MiR-34a is a p53 target that induces apoptosis and G1-arrest. Cell Cycle, 2007; 6(13):1586–1593; doi: 10.4161/cc.6.13.4436

184.

Tasdelen

, Sen

. A hybrid CNN-LSTM model for pre-miRNA classification. Sci Rep, 2021; 11(1):14125; doi: 10.1038/s41598-021-93656-0

185.

Thatai

AKS

, Ammankallu

, Devasahayam Arokia Balaya

, et al. VirhostlncR: A comprehensive database to explore lncRNAs and their targets in viral infections. Comput Biol Med, 2023; 164:107279; doi: 10.1016/j.compbiomed.2023.107279

186.

Thomas

, Thomas

, Sael

. DP-miRNA: An Improved Prediction of Precursor microRNA Using Deep Learning Model. In: IEEE International Conference on Big Data and Smart Computing (BigComp). 2017; 1:96–99. IEEE Xplore, Piscataway, NJ, USA. doi: 10.1109/BIGCOMP.2017.7881722

187.

Thomson

, Dinger

. Endogenous microRNA sponges: Evidence and controversy. Nat Rev Genet, 2016; 17(5):272–283; doi: 10.1038/nrg.2016.20

188.

Tong

, Cui

, Wang

, et al. TransmiR v2.0: An updated transcription factor-microRNA regulation database. Nucleic Acids Res, 2019; 47(D1):D253–D258; doi: 10.1093/nar/gky1023

189.

Tran Vdu

, Tempel

, Zerath

, et al. miRBoost: Boosting support vector machines for microRNA precursor classification. RNA, 2015; 21(5):775–785; doi: 10.1261/rna.043612.113

190.

Wang

, Lu

, Miao

, et al. Cepred: Predicting the co-expression patterns of the human intronic microRNAs with their host genes. PLoS One, 2009a;4(2):e4421; doi: 10.1371/journal.pone.0004421

191.

Wang

, Li

, Yue

, et al. NMCMDA: Neural multicategory MiRNA-disease association prediction. Brief Bioinform, 2021; 22(5):bbab074; doi: 10.1093/bib/bbab074

192.

Wang

, Guan

, Khan

, et al. LMI-DForest: A deep forest model towards the prediction of lncRNA-miRNA interactions. Comput Biol Chem, 2020; 89:107406; doi: 10.1016/j.compbiolchem.2020.107406

193.

Wang

, Zhang

, Sun

, et al. Predicting the potential human lncRNA-miRNA interactions based on graph convolution network with conditional random field. Brief Bioinform, 2022a;23(6):bbac463; doi: 10.1093/bib/bbac463

194.

Wang

, El Naqa

. Prediction of both conserved and nonconserved microRNA targets in animals. Bioinformatics, 2008; 24(3):325–332; doi: 10.1093/bioinformatics/btm595

195.

Wang

, Gao

, Wang

, et al. miProBERT: Identification of microRNA promoters based on the pre-trained model BERT. Brief Bioinform, 2023; 24(3):bbad093; doi: 10.1093/bib/bbad093

196.

Wang

, Xuan

, Zhao

, et al. High-resolution human core-promoter prediction with CoreBoost_HM. Genome Res, 2009b;19(2):266–275; doi: 10.1101/gr.081638.108

197.

Wang

, Zhang

, Li

, et al. MicroRNA identification based on sequence and structure alignment. Bioinformatics, 2005; 21(18):3610–3614; doi: 10.1093/bioinformatics/bti562

198.

Wang

, Yu

, Li

, et al. KGDCMI: A new approach for predicting circRNA-miRNA interactions from multi-source information extraction and deep learning. Front Genet, 2022b;13:958096; doi: 10.3389/fgene.2022.958096

199.

Weis

, Reinberg

. Transcription by RNA polymerase II: Initiator-directed formation of transcription-competent complexes. FASEB J, 1992; 6(14):3300–3309; doi: 10.1096/fasebj.6.14.1426767

200.

Wen

, Cong

, Zhang

, et al. DeepMirTar: A deep-learning approach for predicting human miRNA targets. Bioinformatics, 2018; 34(22):3781–3787; doi: 10.1093/bioinformatics/bty424

201.

Wong

, Wang

, You

, et al. GKLOMLI: A link prediction model for inferring miRNA-lncRNA interactions by using Gaussian kernel-based method on network profile and linear optimization algorithm. BMC Bioinformatics, 2023; 24(1):188; doi: 10.1186/s12859-023-05309-w

202.

Xia

, Feng

, Lei

, et al. Comprehensive characterization of tissue-specific circular RNAs in the human and mouse genomes. Brief Bioinform, 2017; 18(6):984–992; doi: 10.1093/bib/bbw081

203.

Xiao

, Zuo

, Cai

, et al. miRecords: An integrated resource for microRNA-target interactions. Nucleic Acids Res, 2009; 37(Database Issue):D105–D110; doi: 10.1093/nar/gkn851

204.

, Chen

, Lu

, et al. SPMLMI: Predicting lncRNA-miRNA interactions in humans using a structural perturbation method. PeerJ, 2021; 9:e11426; doi: 10.7717/peerj.11426

205.

Xuan

, Guo

, Huang

, et al. MaturePred: Efficient identification of microRNAs within novel plant pre-miRNAs. PLoS One, 2011; 6(11):e27422; doi: 10.1371/journal.pone.0027422

206.

Xue

, Li

, He

, et al. Classification of real and pseudo microRNA precursors using local structure-sequence features and support vector machine. BMC Bioinformatics, 2005; 6:310; doi: 10.1186/1471-2105-6-310

207.

Yan

, Wu

, Wang

, et al. PESM: Predicting the essentiality of miRNAs based on gradient boosting machines and sequences. BMC Bioinformatics, 2020; 21(1):111; doi: 10.1186/s12859-020-3426-9

208.

Yan

, Chao

, Tu

, et al. Improving the prediction of human microRNA target genes by using ensemble algorithm. FEBS Lett, 2007; 581(8):1587–1593; doi: 10.1016/j.febslet.2007.03.022

209.

Yang

, Shao

, Bofill-De Ros

, et al. AGO-bound mature miRNAs are oligouridylated by TUTs and subsequently degraded by DIS3L2. Nat Commun, 2020a;11(1):2765; doi: 10.1038/s41467-020-16533-w

210.

Yang

, Wang

, Lin

, et al. LncMirNet: Predicting lncRNA-miRNA interaction based on deep learning of ribonucleic acid sequences. Molecules, 2020b;25(19):4372; doi: 10.3390/molecules25194372

211.

Yang

, Song

. A stacked autoencoder-based miRNA regulatory module detection framework. Int J Comput Intell Syst, 2019; 12(2):822–832

212.

Yang

, Wan

. Identification of MicroRNA regulatory modules by clustering MicroRNA-target interactions. IEEE Access, 2020; 8:154133–154142

213.

Yang

, Wang

, Li

. MiRTif: A support vector machine-based microRNA target interaction filter. BMC Bioinformatics, 2008; 9(Suppl 12):S4; doi: 10.1186/1471-2105-9-S12-S4

214.

Yang

, Wu

, Wang

, et al. dbDEMC 2.0: Updated database of differentially expressed miRNAs in human cancers. Nucleic Acids Res, 2017; 45(D1):D812–D818; doi: 10.1093/nar/gkw1079

215.

Yao

, Nong

, Qin

, et al. Identifying circRNA-miRNA interaction based on multi-biological interaction fusion. Front Microbiol, 2022; 13:987930; doi: 10.3389/fmicb.2022.987930

216.

, Qin

, Macara

, et al. Exportin-5 mediates the nuclear export of pre-microRNAs and short hairpin RNAs. Genes Dev, 2003; 17(24):3011–3016; doi: 10.1101/gad.1158803

217.

Yones

, Stegmayer

, Milone

. Genome-wide pre-miRNA discovery from few labeled examples. Bioinformatics, 2018; 34(4):541–549; doi: 10.1093/bioinformatics/btx612

218.

Yoon

, Abdelmohsen

, Gorospe

. Posttranscriptional gene regulation by long noncoding RNA. J Mol Biol, 2013; 425(19):3723–3730; doi: 10.1016/j.jmb.2012.11.024

219.

Yoon

, Abdelmohsen

, Gorospe

. Functional interactions among microRNAs and long noncoding RNAs. Semin Cell Dev Biol, 2014; 34:9–14; doi: 10.1016/j.semcdb.2014.05.015

220.

Yousef

, Abdallah

, Allmer

. maTE: Discovering expressed interactions between microRNAs and their targets. Bioinformatics, 2019; 35(20):4020–4028; doi: 10.1093/bioinformatics/btz204

221.

Yousef

, Goy

, Bakir-Gungor

. miRModuleNet: Detecting miRNA-mRNA regulatory modules. Front Genet, 2022; 13:767455; doi: 10.3389/fgene.2022.767455

222.

Yousef

, Goy

, Mitra

, et al. miRcorrNet: Machine learning-based integration of miRNA and mRNA expression profiles, combined with feature grouping and ranking. PeerJ, 2021; 9:e11458; doi: 10.7717/peerj.11458

223.

Yousef

, Jung

, Kossenkov

, et al. Naive Bayes for microRNA target predictions—Machine learning for microRNA targets. Bioinformatics, 2007; 23(22):2987–2992; doi: 10.1093/bioinformatics/btm484

224.

, Wang

, Li

, et al. SGCNCMI: A new model combining multi-modal information to predict circRNA-related miRNAs, diseases and genes. Biology (Basel), 2022a;11(9):1350; doi: 10.3390/biology11091350

225.

, Jiang

, Jin

, et al. preMLI: A pre-trained method to uncover microRNA-lncRNA potential interactions. Brief Bioinform, 2022b;23(1):bbab470; doi: 10.1093/bib/bbab470

226.

Zeng

, Zhu

, Yan

. Towards accurate human promoter recognition: A review of currently used sequence features and classification methods. Brief Bioinform, 2009; 10(5):498–508; doi: 10.1093/bib/bbp027

227.

Zeng

, Cullen

. Structural requirements for pre-microRNA binding and nuclear export by Exportin 5. Nucleic Acids Res, 2004; 32(16):4776–4785; doi: 10.1093/nar/gkh824

228.

Zhang

, Chen

, Yin

. Prediction of potential miRNA-disease associations through a novel unsupervised deep learning framework with variational autoencoder. Cells, 2019a;8(9):1040; doi: 10.3390/cells8091040

229.

Zhang

, Liu

, Chen

, et al. Predicting lncRNA-miRNA interactions based on interactome network and graphlet interaction. Genomics, 2021; 113(3):874–880; doi: 10.1016/j.ygeno.2021.02.002

230.

Zhang

, Meng

, Luan

, et al. Plant miRNA-lncRNA interaction prediction with the ensemble of CNN and IndRNN. Interdiscip Sci, 2020; 12(1):82–89; doi: 10.1007/s12539-019-00351-w

231.

Zhang

, Tang

, Zhou

, et al. LncRNA-miRNA interaction prediction through sequence-derived linear neighborhood propagation method with information combination. BMC Genomics, 2019b;20(Suppl 11):946; doi: 10.1186/s12864-019-6284-y

232.

Zhang

, Zhang

, Chen

, et al. Circular intronic long noncoding RNAs. Mol Cell, 2013; 51(6):792–806; doi: 10.1016/j.molcel.2013.08.017

233.

Zhao

, Qiu

, Zhou

, et al. Graph embedding ensemble methods based on the heterogeneous network for lncRNA-miRNA interaction prediction. BMC Genomics, 2020; 21(Suppl 13):867; doi: 10.1186/s12864-020-07238-x

234.

Zhao

, Zhang

, Chen

, et al. Long noncoding RNAs: Fine-tuners hidden in the cancer signaling network. Cell Death Discov, 2021; 7(1):283; doi: 10.1038/s41420-021-00678-8

235.

Zheng

, You

, Wang

, et al. MLMDA: A machine learning approach to predict and validate MicroRNA-disease associations by integrating of heterogenous information sources. J Transl Med, 2019a;17(1):260; doi: 10.1186/s12967-019-2009-x

236.

Zheng

, Xu

, Zhang

, et al. Nucleotide-level convolutional neural networks for pre-miRNA classification. Sci Rep, 2019b;9(1):628; doi: 10.1038/s41598-018-36946-4

237.

Zhou

, Ruan

, Wang

, et al. Characterization and identification of microRNA core promoters in four model species. PLoS Comput Biol, 2007; 3(3):e37; doi: 10.1371/journal.pcbi.0030037