Abstract
Implementations of Internet of Things (IoT) and automation have already started making their way into the biotechnology industry, but remain restricted to well-funded pharmaceutical companies and laboratories. Universities across the world have been leading innovators in this field and yet a large proportion of academic laboratories are primitive in design. This is due mainly to the investment requirements to execute such upgrades, as well as the costs of training to use high-end infrastructure. This work aims to provide basic IoT solutions for biotechnology labs at a university level. This paper provides detailed insights to several problems laboratories face around the world, including automatic storage of equipment-generated data, pipeline leakages, sensor-related damage of equipment, maintenance of biosafety lab conditions, gas level estimation and basic administrative issues such as logging of equipment (samples, pipette locations, reagents, machinery status, etc.). We attempt to compile the applications of IoT to address these issues while considering the financial constraints of academic laboratories. We discuss the component requirements and approximate cost of implementation for solutions that aim to minimize human errors, thus enabling the reproducibility of results. Furthermore, current developments and future research directions were put forth toward cutting-edge computational applications of machine learning and artificial intelligence in the biotechnology.
Introduction
The biotechnology industry has always struggled with reproducibility of results. 1 Applications of computer science such as Internet of Things (IoT), automation, cloud computing, and artificial intelligence (AI) can be the key to standardizing protocols and minimizing human error. 2 Experiment automation can reduce human error, reduce contamination and speed up the pace of innovations. Automation will also lead to improvement in the overall efficiency of the development processes.
Cloud computing-based bioinformatics software like DNAnexus deals with the management of high volumes of data. It also provides platforms for integrating genetic and other types of data for global collaborations and helps to analyze large volumes of data generated by laboratories globally. AI is being developed to supplement the drug design process by running simulations as well as 3D structure prediction tools like Alfa folds. Based on the chemical environment of the targets, it also predicts the effect of the compounds on targets and the safety measures required before synthesizing or producing these compounds. 3
IoT has revolutionized many avenues of biotechnology, including precision farming techniques in agricultural biotechnology such as Ecovative, where plant-based meat, biodegradable packaging, and leather-like textiles are created using their Mycelium foundry through partnerships and collaborations. And smart wearables obtain important biomedical data such as Withings BPM Connect, a wearable armband, which measures the user's blood pressure with immediate on-screen results in a color-coded format. Apps like Health Mate app connect users so they can share their results with doctors as well.
IoT-Based Automation in Life Sciences
TetraScience has developed cutting-edge IoT-based lab solutions. Here, the primary goal is to integrate a diverse range of lab equipment and associated data collection software into a single platform. A large portion of life science experiments are done manually and can be very time-consuming; TetraScience works to automate these processes. The company collaborated with Biogen to automate their cell counting procedures, which used to be a cumbersome task, produced data in different formats, and suffered from inherent data integrity risks.
The TetraScience platform streamlined the movement of the data from the source (cell counter and PC) to targets (Electronic laboratory notebooks) in a harmonized manner. The TetraScience platform also resolves bioprocessing problems identifying purification techniques, scouting flow rates, 4 and alerting users if there is any equipment that is malfunctioning—potentially saving the experiment and hardware integrity. TetraScience also works on solving administrative issues like tracking the current state of equipment (out of service, in use, maintenance requirement) and associated scientists. This type of data is usually being recorded manually in laboratory records, calendars, diaries, etc., at university-level life science labs. 5
Emerald Cloud Lab has also developed a similar solution. Its integrated platform stores all experimental data and possesses administrative capabilities. 6 This company has also developed a laboratory set up over the cloud, meaning a researcher could perform experiments without ever having to leave the desk. Scientists could perform experiments and receive data from anywhere in the world, with an active internet connection. The system gives full control to the user to determine the experimental conditions. The platform contains numerous functions for data analysis, visualization, etc. It also has 3,500 data processing and manipulation functions in a proprietary language. Therefore, this platform improves the reproducibility of results, reduces cost, and makes scientific research widely accessible.
Ginkgo Bioworks is another example of a company using cutting-edge technology and some help from nature to generate optimized microorganisms using genetic engineering and synthetic biology. Their work includes making vaccines that can detect antibacterial activity, thus contributing to avoiding antibiotic resistance, which leads to a significant number of deaths every year. Their foundries and codebase are very advanced and enable their labs to be highly automated and require very little human interaction during daily activities and experiments. 7 Whenever a piece of equipment is missing or out of order, the system automatically places an order for the piece or calls for maintenance.
Elemental Machines is another company working on creating the next generation of IoT-enabled laboratories around the world. Elemental Machines has developed proprietary sensors (Element A, Element D, and element T) that can be attached to different types of machines. These sensors record real-time data generated in their platform and send alerts on mobile phones if the conditions maintained are not optimal. These sensors also measure changes in environmental conditions that may indirectly lead to the malfunctioning of other machines. For example, when humidity drops, several machines such as the liquid handling systems might malfunction. Element A sensor measures humidity, pressure, temperature, and light levels of the environment. Element D sensor measures data collected from other instruments like incubators, balances, freezers, etc., and automatically stores it on the platform. Element T sensor measures temperatures; it is usually used in freezers that operate at very low temperatures.
Another company combining IoT with biotechnology is LabGenius. 8 They are making huge strides towards the automation of gene synthesis technology and the creation of novel compounds for new biological solutions. Their primary focus is on hit and lead generation and making use of machine learning to implement these processes and reduce failure rate. They focus on important factors like governing drug development, including the specificity so that it only affects the specific target; expression levels, which indicates the yield levels of the protein being made in the cell factories; and stability, which affects shelf life, drug potency and immunogenicity. LabGenius is one of the pioneering companies combining IoT and biotechnology.
Automated liquid handling systems are another innovation that provides automatic dispensing of liquids. This involves automated pipetting systems and microplate washers. Automated liquid handling systems remove the need for a lot of manual labor for processes like sequencing, bioassays, and high-throughput screening. Liquid-handling robots contain a control center, which controls movement, a washing station, liquid-dispensing heads, and a wide variety of sensors to control the amount of sample dispensed. These sensors also provide information to the control center to orchestrate accurate experiments. These systems are sometimes also integrated with liquid chromatographic systems, heating/cooling systems, shakers etc. Such liquid handling machinery has applications in genetics and pharmaceutical research.
Solution4Labs is another company working toward automation using IoT. Its solution consists of LIMS systems that integrate everything into an ecosystem of digital data. Solution4Labs creates labs for their users certified to work with standards like 21 CFR Part 11, GAMP, and ISO 17025, and also provides access to analysts based on their qualification levels by restricting access to unneeded equipment. It also provides barcodes or RFID codes for quickly locating samples, down to the shelves they are kept on, as well as instruction for analysts to comply with for their systems by guiding them through all steps of the procedure and full history.
Solution4Labs helps in resource management by controlling reagent consumption, monitoring inventory levels, and scheduling servicing, calibration and validation of lab equipment. It also maintains an individual history of the same for each instrument. Finally, it makes sure that integrated instruments send data straight to the software, where it can be viewed, analyzed and computed using online calculations. This is done by its SDMS data management system while capturing of metadata and raw data from instruments can be done using Data Manager software.
The Current State of Non-IoT Based Life Science Laboratories
CLOUD-BASED LOGGING OF EQUIPMENT AND READING FLUCTUATIONS
University-level laboratories consist of a wide variety of equipment that generates certain types of reading that can be manually noted (weight, balance, incubator) or stored on equipment-specific software (HPLC, spectrophotometer, PCR). This is innately error-prone and leads to unnecessary delays in experimental progress. There is no integrated platform that stores readings from all kinds of equipment and also exports them into electronic lab notebooks. General administrative problems like labelling samples, tracking equipment details, locating reagents, etc., are still tracked using heavy notebooks. Damages in hardware lead to fluctuations in readings which can also delay experiments. Fluctuations in freezers cause large problems as it stores samples that have taken weeks and months to develop by the scientists. Overheating of certain devices can also cause variations in other equipment and contaminate other device's readings. 9
GAS LEAK IN THE PIPELINE AND GAS LEVEL IN CYLINDERS
In a biotechnology lab, Gas Chromatography and Evaporative Light Scattering Detector (ELSD) use a multitude of gasses like Helium, Hydrogen, Nitrogen, Hydrogen Disulphide etc. for various experiments. The currently available options for monitoring the gas leakage are in-built devices while there is a need to address the real-time monitoring of the pipeline attachment that is being used when the cylinders are fitted at a distance. During the course of time and the open environment of the pipeline, joints may get loose over time and the chances of gas leakage may considerably increase at those particular points. There is a need for locating the exact location of the gas leaks on an integrated platform at the right time because these pipelines are quite long.
The various gasses such as Hydrogen, Nitrogen, Carbon Dioxide, Hydrogen Sulfide, Helium, etc. are used constantly for various experiments going on simultaneously in biotechnology labs which means they get depleted very quickly, and the experiments require a regular supply of such gasses. The levels need to be monitored since they are not easily available everywhere. Cylinders take a long time to be delivered and especially during the covid period, they take even longer, and it becomes a cumbersome task to source them. This is why gas levels in each cylinder need to be monitored cautiously so that new cylinders can be ordered timely, without disrupting the workflow of the lab. Manual observation by the gas pressure gauge is a difficult task to visualize the pressure periodically, while IoT-based sensors can do that work without even coming to the lab and can be monitored from the office or home.
BIOSAFETY ROOM CONTROL AND ADMINISTRATIVE ISSUES
There are 4 levels of Biosafety rooms. Maintaining strict conditions at level 3 and 4 is crucial to prevent harmful contamination. There must be auto locking doors and a controlled air ventilation system to make sure proper airflow direction is maintained. The air must be HEPA filtered, conditioned and recirculated within the laboratory. In level 4, biosafety laboratory there must be a minimum of two auto locking doors before entering safety cabinets. There must be entry of trained and authorized personnel only. Specific climate control must be maintained within these laboratory (temperature, pressure, humidity etc.). A negative pressure must be maintained in these labs.
Usually, logging of equipment becomes a problem in larger biotechnology laboratories as the reagents, salts, apparatus, etc. are in huge numbers. Whether something is finished and needs to be ordered again, or something is damaged or moved to another lab, all this requires tedious logging time which creates delay in the workflow of the lab. It enhances the possibility of mistakes in the logging system due to human error causing problems and conflict inside the laboratory. Recording readings, labeling samples, tracking reagents etc., can all be categorized under administrative problems and be stored in electronic laboratory notebooks. 10
IoT Solutions for Common Problems Faced in Life Science Labs
GAS LEVEL ESTIMATION WITH LOAD CELL SENSOR
The proposed model takes care of estimating the usage of various gases stored in their respective cylinders and issuing alerts of replacement in advance of the cylinders depleting. The model involves a load cell sensor placed underneath each cylinder, which would estimate the usage of a gas by measuring the fraction of deficit in the weight of the filled cylinder and an empty one. 11 The readings taken by the sensor would be then processed by a microcontroller and the level of remaining gas would be displayed in a web app on a desktop or a mobile application. 12 At the same time, as per the rate at which the gas is being used, a threshold value of the gas is fixed, if the weight drops below this threshold, then a booking alert is issued, to ensure we have a replacement cylinder in time before the current one depletes and avoid hindrance to any ongoing experiments. 13
Implementation
A circuit can be implemented (Fig. 1) with the primary component being a class C3 load cell sensor that gets the readings of the weight of the gas cylinder. The sensor is used in conjunction with an ATmega328p based microcontroller such as an Arduino Uno to read the continuous data of measured weight and determine the percentage of gas in the cylinder remaining as per our calculations. 14 We attach a Wi-Fi module such as the ESP8266 with the Arduino then so that we can transmit the data of the level of gas remaining in the cylinder over Wi-Fi and display the results on a web app or a mobile application. 15 The aim of this implementation was to make it cost efficient, describes approximate prices of the hardware components required.
Circuit diagram implementing the aforementioned model for gas level estimation in cylinders.
FUZZY INFERENCE SYSTEM FOR FLUCTUATION
A model for combating the fluctuations in temperature of crucial lab equipment such as cold storage refrigerators and incubators. Control systems with fuzzy logic are a reliable method of keeping track of any temperature fluctuations which might arise due to several issues such as voltage undershoot or overshoot. We install a microcontroller-based circuit using fuzzy logic which accurately measures temperature readings and issues an alert if there is a difference in the temperature setting of our equipment and the current measurement (beyond the allowed error approximation). This allows remote monitoring of such equipment which is running in laboratories 24x7. 16
Implementation
We can implement a circuit based on fuzzy logic (Fig. 2) comprising of a highly accurate industrial grade temperature sensor, working in conjunction with a microprocessor such as an Atmega328p (Arduino Uno) to read the continuous data of measured temperature from the equipment and determine if there is any fluctuation outside of the allowed error range. A Wi-Fi module such as the ESP8266 is attached to the microprocessor to transmit this data over Wi-Fi and display the results on a web app or a mobile application. The table below describes the approximate prices for the hardware components.
Circuit diagram of a fuzzy inference system for detecting fluctuation in temperatures.
RFID-BASED TAGGING SYSTEM
We can implement a solution via the use of RFID technology. This consists of RFID tags that come in various shapes and forms such as key cards, to even stickers that can be attached to most of the lab apparatus. The second component of this technology involved an RFID scanner that is used to identify the particular apparatus when scanned. All RFID tags have a certain unique ID, this allows us to allot all the various apparatus a unique number so that when scanned we get the details along with grouping the various apparatus under different categories. 17 Apart from this, the most useful function is that the scanner would be attached to a centralized web app over Wi-Fi where one can keep track of all the various apparatus in use and which are available for use. When a certain apparatus is to be used this system can simply be scanned to change its status as occupied and when the work is done, it can be scanned again to make it available to anyone else. 18
Implementation
We start with tagging the relevant apparatus with a UHF RFID tag and then use a compatible RFID scanner such as the RC522 in conjunction with an Atmega328p based microcontroller such as Arduino Uno (Fig. 3). The apparatus is initially tagged and given a serial number as well as categorized in accordance with their unique ID given by the manufacturer. Thereafter we program our microprocessor such that whenever the RFID scanner scans any apparatus, it lists out its details and availability. A Wi-Fi module such as the ESP8266 is also attached to the microprocessor to transmit this data over Wi-Fi and display the relevant data such as apparatus details and availability on a web app or a mobile application.
Circuit diagram depicting an RFID tagging system, showing an instance of how the scanning dock may be wired for implementation.
For larger apparatus being difficult to move around a portable, handheld version of the circuit can be made, powered via rechargeable li-ion batteries. The table below states the list of hardware components required for this implementation and their approximate prices.
List of Components Required for the Solutions Mentioned Above and Their Approximate Prices
GAS LEAK DETECTION IN THE PIPELINE WITH FBG SENSOR
To ensure the highest accuracy and reliability we start with a pair of FBG pressure sensors and have them connected to a fiber grating demodulator. The demodulator can then be connected to a capable microcontroller such as a raspberry pi 4 to interpret the data from the demodulator and if any leak is detected then issue a critical alert. If the pipeline is transferring flammable gasses, we can also have a fail-safe mechanism to cut off the supply to avoid any mishaps. 19 Since we are using FBG pressure sensors, we will also have the capability of estimating the region of the pipeline where the leak might be, so it is easily located for repair. 20
Implementation
Firstly, FBG pressure sensors are to be installed at two ends of the pipeline coming to the laboratory. Next, the sensors are connected amongst themselves and to a fiber grating demodulator. The readings from the sensors are interpreted by the demodulator and the results are subsequently passed to our microcontroller, which has been programmed to issue any alerts in case of leaks and estimate the location along the pipeline, and also cut off the supply if any flammable gas/liquid is in the pipeline. We also transmit our results over Wi-Fi to see the live statistics of pressure readings of the pipeline on a web app or a mobile application and also see a pictorial representation of the location of the leakage.
CLOUD-BASED LOGGING OF EQUIPMENT GENERATED READINGS
Most of the biotechnology labs have equipment that generates readings and exports them to an external system. We introduce the integration of a cloud platform to take these readings, pass them through custom software that formats the data to be stored in a format we want, and create log files that are continuously updated and stored on the cloud to ensure continuous availability and avoid any loss of data. 21 A lot of the machines have their own software to analyze and store data. Thus, an integrated platform to store readings of all devices and their transfer into individual electronic lab notebooks will streamline the research process. The readings can also be accessed remotely to check the status of experiments and adjustments can thus be made.
Implementation
We scrape the readings coming into the system connected to the equipment and export the data log files in regular intervals as CSV or other relevant format files and store them in the cloud platform of choice, which makes the data remotely accessible from anywhere and also ensures redundancy. 22 The cloud platform also gives us access to virtual machines for processing the collected data without the need to set up additional computing resources in the lab. 23
BIOSAFETY ROOM TEMPERATURE AND ACCESS CONTROL
There is a need for strict temperature and pressure regulation in biosafety rooms at level 3 and level 4. Also, there is a need for strict regulation to allow only authorized personnel in the restricted biosafety rooms to avoid contamination and outbreaks. The prior objective is achieved by placing high accuracy temperature and pressure sensors throughout the biosafety room and connecting it to a microcontroller that can continuously monitor the readings. The access control to the lab can be kept in check by installing an RFID system, in which all authorized personnel have a designated key card that can identify them upon their entry to the room and also maintain logs.
Implementation
First, we can take a high accuracy temperature and humidity sensor and pair it along with a high accuracy pressure sensor, connecting them to an Atmega328p based microcontroller such as Arduino Uno. The temperature sensor to be used can be DHT22 with a measurement range of -40 to +80°C with an accuracy of
Furthermore, for maintaining access control to the biosafety room, we can install an RFID system, consisting of an RFID reader such as an RC522 which will connect to our microcontroller. A number of RFID key cards can then be distributed to authorized personnel that they can use to unlock the biosafety rooms. Since all cards have a unique ID, the microcontroller can be programmed to identify who is entering the room and when. Lastly, a Wi-Fi module such as an ESP8266 is attached to the microcontroller so that all data can be transmitted over Wi-Fi to a centralized web or mobile application that can display constant temperature, humidity and pressure reading, issuing an immediate alert in case of any fluctuations. The web app can also display a log of all the people who have entered the biosafety room on a particular day, along with the time of entry and exit.
IoT based implementations in the laboratories, make the data collection convenient, reproducible and error free. The use of data handling is implemented by collecting the readings from the sensors are recorded by the micro controllers. Consecutively, the collected data is transmitted over Wi-Fi to a centralised web app where they the data is permanently stored for future use
Future Prospects
Machine Learning (ML) is an interesting avenue when we talk about the prospects of biotechnology 24 . This is because using ML and AI opens up faster and more reliable methods for specific processes like drug discovery or vaccine development, helping us tackle problems like an immune response to drugs as well as vaccine efficiency in the body without needing to test physically first. This will save a lot of time as well as be cost-efficient for companies. Also, this would be a big step towards the creation of personalized medicines for specific patients and help with more effective treatment of patients along with other benefits like reduced costs and time-saving. 25 Further research in this space would improve the existing technology and automate repetitive tasks. Also leading to the reduction in human error which means that tasks are completed earlier and more accurately.
AI applications in biotech include drug target identification, drug screening, image screening, predictive modeling, and as a means to comb through the scientific literature and manage clinical trial data. 26 With the amount of data available to biotechnology scientists worldwide, it becomes crucial to rely on AI and ML so that they can parse through the huge data lakes, carry out the data analysis, and store it securely. By leveraging machine learning, AI can manage disparate clinical trial datasets, enable virtual screening and streamline experiments. 27 Besides reducing clinical trial costs, AI can gain otherwise unobtainable insights and feed them back into the drug development process. 28
AI technologies that serve the biotech industry are being developed by several companies and are rapidly becoming indispensable as older methods like classical statistical analysis or manual image scanning reach their practical limits. Like many other disruptive technologies in pharmaceutical research—technologies such as CRISPR gene editing, proteolysis targeting chimera, induced protein degradation and RNA interference, AI has a lot of potential. 29
Integration of AI and ML is essential for players in the pharmaceuticals market today as these will help you gain an advantage over competitors as they can help with reducing the time-to-market as well as improving reliability and consistency of results and all this will help to future-proof against uncertainty and adversity. Even other tasks like analysis of recorded data and any computing required on it can be handled with more efficiency and will take less time as compared to when done by their human counterparts.
Companies like Atomwise which uses machine learning-based discovery engines which combine convolutional neural networks with large chemical libraries for discovery of small molecule medicines where they are screening trillions of compounds in silico overcoming the problem of false positives by creating accurate models and increasing global model predictability through continued generation as well as integration of training data drives and a whole team of scientists evolving ML algorithms for generation of more robust data sets. Agilent Technologies is partnering up with an AI software company for improving the quality and standardization of tissue diagnostics and pathology labs to accelerate diagnostics. Gilead Sciences is using an AI platform for discovery and development of treatments for Nonalcoholic steatohepatitis by introduction of data-driven discovery and development to transform both discovery and delivery of treatment to the patients.
Conclusion
IoT applications have been revolutionizing a wide range of industries, healthcare being an important one. Wearables have been developed that track biomedical information for doctors to have a constant check-up on their patient has been one of the biggest developments. 30 IoT-based biomedical machines will improve innovation in this industry. IoT can help to maintain constant conditions of growth for plants which will improve the agricultural industry. 31 Pharmaceutical companies already have highly automated labs and AI-based drug design software. The implementation of these solutions can be the first step towards automating biotechnology labs at universities across the world which have historically been the birthplace of a wide range of innovations. 32 All future scientists must keep up with the data revolution and harness its abilities.
This article deals with designing an IoT-enabled biotech lab while considering the financial constraints of universities around the world, keeping a simple design to make it user-friendly while preserving scientific creativity in students and scientists. 33 Common problems like administrative issues, hardware damage, basic administration, Biosafety room control are outlined in this article and might be implemented in academic laboratories to improve overall efficiency. Applications of ML and AI will be the future of this field and soon would become financially feasible. With the development of technology, cost of such infrastructure will be reduced and in laboratories around the world, equipment will be communicating among each other and spending hours in the laboratory to perform an experiment will be a task of the past. Outsourcing experiments to other labs will free up time and let scientists focus on solving pressing problems. 34 Entering data both experimental and administrative into physical notebooks will also become obsolete. 35
Firms in the Biotechnology sector must make investments in bioinformatics-based software and universities should train graduates to unlock the true potential of this field. 36 The next generation of scientists must be equipped with basic computational expertise to expand their range of understanding.
Footnotes
Acknowledgments
Authors are grateful to Bennett University, Greater Noida, for providing the providential opportunity and desired assistance.
Author Contributions
Mehul Bhatt and Aman Vazirani contributed equally towards the preparation of the manuscript. Sumant Srivastava contributed toward the preparation of the manuscript. Sarika Chaudhary supervised personnel and contributed towards the preparation of the manuscript.
Author Disclosure Statement
No competing financial interests exist.
Funding Information
Dr. Chaudhary would like to acknowledge DBT grant No. BT/PR28766/BRB/10/1701/2018.