Artificial Intelligence-Based Data-Driven Strategy to Accelerate Research, Development, and Clinical Trials of COVID Vaccine

Sharma, Ashwani; Virmani, Tarun; Pathak, Vipluv; Sharma, Anjali; Pathak, Kamla; Kumar, Girish; Pathak, Devender

doi:https://doi.org/10.1155/2022/7205241

BioMed Research International

On this page

Abstract Introduction Conclusion Data Availability Conflicts of Interest References Copyright Related Articles

Review Article | Open Access

Volume 2022 | Article ID 7205241 | https://doi.org/10.1155/2022/7205241

Artificial Intelligence-Based Data-Driven Strategy to Accelerate Research, Development, and Clinical Trials of COVID Vaccine

Ashwani Sharma,¹Tarun Virmani,¹Vipluv Pathak,²Anjali Sharma,³Kamla Pathak,⁴Girish Kumar,¹and Devender Pathak⁵

Academic Editor: Usman Ali Ashfaq

Received12 Jan 2022

Accepted15 Jun 2022

Published06 Jul 2022

Abstract

The global COVID-19 (coronavirus disease 2019) pandemic, which was caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has resulted in a significant loss of human life around the world. The SARS-CoV-2 has caused significant problems to medical systems and healthcare facilities due to its unexpected global expansion. Despite all of the efforts, developing effective treatments, diagnostic techniques, and vaccinations for this unique virus is a top priority and takes a long time. However, the foremost step in vaccine development is to identify possible antigens for a vaccine. The traditional method was time taking, but after the breakthrough technology of reverse vaccinology (RV) was introduced in 2000, it drastically lowers the time needed to detect antigens ranging from 5–15 years to 1–2 years. The different RV tools work based on machine learning (ML) and artificial intelligence (AI). Models based on AI and ML have shown promising solutions in accelerating the discovery and optimization of new antivirals or effective vaccine candidates. In the present scenario, AI has been extensively used for drug and vaccine research against SARS-COV-2 therapy discovery. This is more useful for the identification of potential existing drugs with inhibitory human coronavirus by using different datasets. The AI tools and computational approaches have led to speedy research and the development of a vaccine to fight against the coronavirus. Therefore, this paper suggests the role of artificial intelligence in the field of clinical trials of vaccines and clinical practices using different tools.

1. Introduction

Virus-caused infectious diseases have long been the most difficult challenge in human health. High-infectivity and high-mortality diseases are particularly feared, and in the past, people viewed them as a tragedy or disasters (Hilleman et al., [1]). Humankind has been able to overcome the irrational dread of death because of advances in recognizing the etiology of viral diseases and knowledge of microbiology, which were followed by the creation of numerous vaccines. Vaccination is often regarded as one of the greatest achievements in medical history. Immunization has saved a lot of lives, and its significance continues to expand. Despite countless efforts to develop qualified and efficient vaccinations, there are inadequate barriers in place to protect populations from diseases that could produce epidemics or pandemics (for example, the Ebola virus epidemic) (Kilbourne, [2]; Gostin et al., [3]). As a result, scientists are working to expand the viral infection that may be prevented by vaccinations, as well as the population groups that will benefit from vaccination in the long term. As of now, Coronaviruses (CoVs) are responsible for the causes of serious illnesses in humans and a variety of animal hosts, including respiratory, gastrointestinal, and systemic diseases. Infections with the CoVs have been found in cattle, swine, rats, cats, mink, dogs, bats, palm civets, horses, camels, ferrets, rabbits, snakes, and a variety of other wild mammals and bird species (Fehr and Perlman [4], Kahn and McIntosh, [5]). Since the very first outbreak in 2002, the Coronaviridae virus family, which causes pneumonia-like symptoms, has been a global danger (Khan et al., [6]). The infections including severe acute respiratory syndrome (SARS) and middle eastern respiratory syndrome (MERS), which first occurred in 2002 and 2013, respectively, caused respiratory and gastrointestinal problems (Hilgenfeld and Peiris, [7]). SARS-COV-2, which has been identified as the virus that causes COVID-19, having symptoms ranging from a common cold to serious respiratory failure, was the source of a third coronavirus episode in 2019 (Kong et al., [8]). Compared to the World Health Organization’s (WHO) declaration of a pandemic, COVID-19 has started spreading and has affected at least 20 million people, with a mortality toll of around 5 lakh at the time of such an assessment (Worldometer, [9]). Due to the inadequacy of lab-based high throughput screening (HTS), virtual screening (VS) has emerged as a preferred tool for discovering effective molecules while hospitals continue to trial and error strategies for COVID-19 drug discovery (Jin et al., [10]; Kandeel and Al-Nazawi, [11]). The strategy of specifically targeting biomolecule (e.g., DNA, protein, RNA, and lipid) by using a computer program, of a cell to suppress its growth and/or activity is known as rational drug discovery which can be recognized by VS (Shoichet, [12]; Lionta et al., [13]). Two significant subgroups of this form of screening are ligand-based and structure-based drug design and discovery (Lionta et al., [13]; Yu and Mackerell, [14]; KeshavarziArshadi et al., [15]). Computationally and experiment-based access can determine the viral protein structures and antiviral candidates can be identified quickly and at a low cost using VS (Senior et al., [16]; Zhang et al., [17]). Furthermore, traditional vaccine development approaches are expensive, and developing an effective vaccine against a specific virus might take many years. COVID-19 vaccines have been developed and manufactured with much effort, and the efforts to advance vaccine clinical trials have been tremendous. Coronaviruses are positively stranded RNA viruses that have their genome packaged into the nucleocapsid (N) protein and are surrounded by the membrane (M), envelope (E), and spike (S) proteins (Li, [18]). While many coronavirus vaccine experiments targeting various structural proteins were done, most of these efforts came to an end soon after the SARS and MERS outbreaks. The first human trial of the mRNA-based vaccination targeting the SARS-CoV-2’s S protein began on March 16, 2020, as an expedient response to the COVID-19 pandemic. S protein is the coronavirus’s most superficial and protrusive protein, and it plays a critical function in the entry of the virus. Because of their capacity to induce neutralizing antibodies that block host cell entry and infection, the full-length S protein and its subunit S1 (which contains the receptor-binding domain) have frequently been employed as vaccine antigens in the development of SARS and MERS vaccines. Current coronavirus vaccines, particularly S protein-based vaccines, may, however, have challenges with producing complete protection and potential safety concerns (Roper and Rehm, [19]; De Wit et al., [20]).

Nonetheless, sterile immunity and full protection are desired outcomes of a COVID-19 vaccine. Furthermore, it is becoming increasingly obvious that diverse immune responses, such as those elicited by cell-mediated or humoral immunity, are more important predictors of protection than antibody titers alone (Ong et al., [21]). One of the technology reverse vaccinology (RV), which is aimed at uncovering viable vaccine candidates through bioinformatics analysis of the pathogen genome, has transformed vaccine research in recent years.

2. Reverse Vaccinology

Reverse vaccinology (RV) was introduced in the early 1990s as a genome-based vaccine design approach (Rappuoli, [22]; Bullock et al., [23]) and attributed to the reason that bacterial culturing was no longer needed for selecting vaccine targets; the field was transformed to a more efficient status (Soria-Guerra et al., [24]; Heinson et al., [25]; Bruno et al., [26]). Its goal is to use bioinformatics to analyze the pathogen genome to find a good vaccination candidate. RV has been used to develop vaccines against infections like Group B meningococcus, which resulted in the approval of the Bexsero vaccine (Folaranmi et al., [27]). The research and application of vaccine-like compounds to humans date back to prehistoric times. Hilleman et al. drew a simple diagram to show this history (Figure 1). We are living in the present era of vaccine development, which is more effective and productive than any previous period in history, according to this diagrammatic overview. This advancement has been achieved due to generous financial support as shown in Figure 1 (Hilleman et al., [1]).

In recent decades and for future perception, AI is the most emerging and demanding scientifically engineered technique. Through this, the computational understanding of machines can be obtained by incorporating intelligent behavior to innovate intelligent and smart machines. AI consists of various pieces of techniques, tools, and algorithms such as neural networks, symbolic AI, deep learning, machine learning, and genetic algorithms. These tools are growing and showing impact in different fields like military, space, robotics, and health. AI term was given by John McCarthy while he was at a conference on this subject in 1956 (McCarthy, [28]; Turing, [29]; Russell et al., [30]). In recent years, the advanced featuring tool of AI is machine learning (ML) reorganized in different fields of engineering and science. Nowadays, it is largely adopted in our daily lives, but the ability to find out the conceptual abstract from the large volume of data and feature learning is the most powerful contribution of ML as a tool of AI (Lecun et al., [31]; Grover and Toghi, [32]; Sun et al., [33]). The subbranches of AI are depicted in Figure 2.

AI is applied to medicine and has two main divisions, which are virtual and physical; the virtual part is acted by ML (also called deep learning), and its representation is achieved by mathematical algorithms, as a count of its experience; it improves learning. Based on ML algorithms, there are three divisions: (1)Supervised (prediction and classification algorithms based on former examples)(2)Unsupervised (patterns finding ability)(3)Reinforcement learning (rewards sequences and punishments are used to build a scheme for operation in a particular problem)

Earlier, AI has raised and is still raising the impact of its techniques in genetics and molecular medicine discoveries due to algorithms of machine learning and management of knowledge. A great example of success in the development of medicine and vaccine is determined by unsupervised protein-protein interaction algorithms, which can lead to remedial target discoveries (Theofilatos et al., [34]). Adaptive evolutionary algorithms and state-of-the-art clustering are the two methods used in the combination and that novel methodology is named “evolutionary enhanced Markov clustering.” More than 5000 protein complexes are under this permitted prediction from which at least one gene ontology function phrase reinforced over 70% of the results. The development of a novel computational methodology is permitted to identify single-nucleotide polymorphisms (SNPs) of DNA variants to predict the traits or diseases by employing revolutionary evolutionary. This works by embedding algorithms that are more robust and less prone to over-fitting problems that occur when a model has too many parameters with the number of observations (Rapakouliaet al., [35]; Theofilatos et al., [34]). According to the predictions, the most effective method for drug development is the Graph Convolution Neural Network (GCNN) (Duvenaud et al., [36]; Kearnes et al., [37]). These networks can retain graphs and extract properties from the information encoded within the compound characteristics. AI contributes to learning which can be successfully done by GCNN for a molecule and compound’s prediction such as property and protein interface estimation (Fout et al., [38]; Liu et al., [39]). Some models are sequence-based like proteomics, transcriptomics, and genomics. They have shown an impact in recent years because of natural language processing (NLP) domain advancement, but more advanced generation models are context-based models which gain the attention from sequence-based models (Devlin et al., [40]).

Furthermore, vaccine development techniques are being adapted to certain countries’ economic and health needs. This tendency has a direct impact on the goods in development as well as the quantity and types of clinical studies conducted. This chapter briefly summarizes the role of AI-based models in COVID-19 drug discovery or research and vaccine development (clinical trials). It is to be proposed that a concentrated effort be made to use AI methodologies to utilize information from preexisting data.

2.1. Coronavirus Structure

Coronavirus’s basic structure is made up of five proteins, which are named as follows: spike proteins (S), Membrane proteins (M), envelope glycoproteins (E), nucleocapsid proteins (N), and hemagglutinin esterase (HE). Spikes are thought to have three heads (S1) and a trimeric stalk, according to several studies (S2). S1 connects to a host cell’s particular surface receptors for viral attachment, whereas S2 binds to the viral and host membranes at the time of virus penetration. This makes it possible for the virus to infect the cells of the host (Li, [18]). The structural proteins are M proteins that aid in the determination of a virus’s shape (Neuman et al., [41]). These are the most common proteins in CoVs. These proteins are primarily responsible for RNA packing (Tang et al., [42]). The protein E is extensively produced inside the infected cell during the replication cycle (Venkatagopalan et al., [43]). Furthermore, E protein is involved in viral budding, assembly, and morphogenesis (Nieto-Torres et al., [44]). Phosphoproteins with the ability to attach to the RNA genome are known as N proteins (de Haan and Rottier, [45]). The N protein is required for virion assembly, replication, and CoVs transcription. It aids in the budding and assembly of viruses (Tooze et al., [46]). Certain enveloped viruses have a glycoprotein called HE. The host cell surface receptor is a sialic acid derivative, and HE aids in its adhesion or recognition. It is the one who is responsible for the receptor’s demise (de Groot, [47]; Zeng et al., [48]).

2.2. COVID-19: Molecular Mechanisms and Target Selection

The spikes on the coronavirus bind to a receptor present on the surface of a cell. After fusing with the cell membrane, the virus releases its RNA genome into the cell. The cell then produces copies of RNA as well as structural proteins required for the assembly of new virus particles, which are then discharged into the body. The role of the immune system is to destroy pathogens such as bacteria and viruses that are causing disease. Its first step is an innate immune response that is to send a variety of weapons to fight infection. But if the pathogens are new to the body and this response does not work to fight infection then, an adaptive immune response comes into action. During the adaptive immune response, specialized cells envelop the virus and deliver antigens, which activate immune cells. The two types of white blood cells are B cells and T cells which play a role in adaptive immunity. Antibodies are specialized proteins that are produced by B cells that bind to pathogens and prevent them from infecting healthy cells. T cells can destroy virus-infected cells that prevent them from replicating the virus. Meanwhile, memory B and T cells record the antigens, ensuring that the body responds rapidly if the coronavirus encounters again (Waltz, [49]). B cells are important in humoral immunity because they can secrete antibodies that neutralize the antigen. The ElliPro web tool (http://tools.iedb.org/ellipro/) was used to forecast both discontinuous and linear B cell epitopes. B cell epitopes assist in the detection of viral infections in the immune system. At the 0.51 threshold, ABCpred (http://crdd.osdd.net/raghava/abcpred/) was utilized to forecast 14-mer B cell epitopes for target proteins (Saha and Raghava, [50]; Ponomarenko et al., [51]; Rafi et al., [52]). T cell epitopes are important in the development of vaccines. It saves money and time as compared to laboratory experiments. 8–11 mer MHC class-I and 11–14 mer MHC class-II epitopes were predicted using the IEDB consensus technique (http://tools.iedb.org/mhcii/) (Zhang et al., [53]; Tahir ulQamar et al., [54]). Figure 3(a) shows a schematic representation of adaptive immunity targeting pathogens for killing.

(a)

(b)

2.3. COVID-19: Vaccines

A vaccine usually contains an agent that looks like a disease-causing germ (microorganism) and is manufactured from weakened or destroyed microbes, one of their surface proteins, or their toxins. The agent induces the body’s immune system to detect the agent as a threat as well as any microbes connected with it (Melief et al., [55]; Bol et al., [56]). Vaccines can be used as preventive or therapeutic measures (Brotherton, [57]; Frazer, [58]). Vaccines are the most significant public health measure against COVID-19 around the world as SARS-CoV-2 is a highly contagious virus infecting people all over the globe (Amanat and Krammer, [59]). Figure 3(b) shows the role of the vaccine in preventing the spread of a virus. COVID-19’s spread shows no signs of halting, and its fatality rate is relatively high when compared to other viral-based infections; the development of vaccines and antiviral treatments against SARS-CoV-2 is very critical and essential. Most vaccines take years to develop ranging from 5 years for Ebola and 40 years for polio. On average, vaccines took 15 years. The vaccine trial process consists of various steps that must be followed quantitatively and systematically. The length of this process is proportional to the vaccine’s purpose and nature, which is to protect healthy people from pathogen infection (Deb et al., [60]; Thanh Le et al., [61]). As this virus is fatal, our societies and economies are unlikely to return to normal until a highly effective vaccination has been given to a large section of the world’s population.

The hunt for a safe and effective vaccine was grown to be a massive project involving thousands of researchers working in laboratories pursuing different platforms for COVID-19 vaccine all over the globe. COVID-19 vaccine development platforms include several novel technologies. Vaccines work by presenting antigens that cause the immune system to respond without getting the individual sick. Vaccines come in a variety of forms, including weakened versions of the complete virus, DNA or RNA, and specific virus fragments that drive a cell to produce a specific virus fragment. The costly and long process of vaccine development can be accelerated using computational methods. However, because of the urgent necessity and deteriorating situation worldwide, some scientists (Pfizer) raced to develop COVID-19 vaccines. The pharmaceutical industry incorporated artificial intelligence in several areas of the development of vaccines and trials during the process. The scientists need to go through each data set for checking errors and other irregularities that occur naturally while collecting millions of data points. Advantageously, technology helped to reduce the effort smoothly. As reported by Cohen in Science magazine [62], the American company Moderna has reduced the time it takes to develop a human-testable vaccine prototype by utilizing bioinformatics technologies in which AI looks to play a crucial role. Moderna was one of the first to introduce a COVID-19 vaccination that was effective. Moderna is also utilizing artificial intelligence to aid in the development of mRNA sequences. Dave Johnson, an AI Officer, and Moderna’s Chief Data show how robotic automation and AI algorithms allowed them to go from manually creating roughly 30 mRNAs per month to being able to manufacture over 1,000 per month. Its use of AI to speed up development was one of the reasons it was able to achieve this breakthrough so swiftly (Gast, [63]). AstraZeneca, a major contributor to the COVID-19 vaccine, is employing artificial intelligence not only in the development of the Covishield vaccine but also in drug discovery. AstraZeneca was one of the first companies to use artificial intelligence in the healthcare sector. To make the drug-making method less expensive, safer, and faster, the company has introduced AI into each step of the research and development process. AstraZeneca has combined knowledge graph and image analysis to get new insights into diseases and detect biomarkers 30% faster than human pathologists (Beatrice, [64]). Thermoregulated storage is required for the entirety of COVID-19 vaccinations. Covishield from Oxford-AstraZeneca and Covaxin from Bharat Biotech, for example, demand a storage temperature of 2–8°C. IoT based on sensor technology, which allows for continuous real-time data monitoring, can help to ensure a reliable storage system. If the temperature changes, the sensors will detect it and issue a device alert for the next vaccination shipment (Kumar and Veer, [65]). AI was successfully employed by Pfizer to run vaccine trials and expedite distribution. Pfizer, on the other hand, used AI throughout the vaccine development process to ensure that the COVID-19 vaccine met the needs of individuals. Pfizer began automating its research and development activities and incorporating artificial intelligence into its working system even before the epidemic. The company employed artificial intelligence algorithms to help identify signals amid millions of data points in its 44,000-person research during the vaccine trials. AI was applied in several aspects of vaccine development and trial during the vaccine development process by the pharmaceutical industry. After satisfying the key efficacy case counts, the data were analyzed and made available in approximately 22 hours with the help of an ML tool, i.e., Smart Data Query (SDQ). Throughout the study, the ML technique assured data quality, requiring very little human interaction (Beatrice, [66]). Table 1 shows the implications of AI/ML in some of the vaccines for SARS-COV-2. Live attenuated, inactivated, and inactivated with adjuvant vaccines are still being developed using the traditional process. Recombinant subunit vaccines as well as more advanced approaches along with DNA and RNA-based vaccines are also being used (Thanh Le et al., [61]; Zhang et al., [17]; Lurie et al., [67]).

3. AI in Vaccine Research

Studying the proteins that make up the virus, such as the spike protein (S), is one of the roles and functions of AI in vaccine development. An AI system can sort through thousands of components in a complicated structure to find the ones most likely to elicit a strong immunological response. To ensure that a vaccine remains effective over time, AI systems must identify components that are unlikely to change or mutate. In the search for a vaccine, a crucial role has been seen for computational analyses and machine learning algorithms. These technologies are helping in assisting researchers in better understanding the virus with its structure and predicting which of its components will elicit an immune response, which is an important and main stage in vaccine development. These techniques can assist researchers in selecting the components for possibly potential vaccines and making sense of experimental data. By merging data from numerous experimental and real-world sources, AI allows scientists to derive insights. They also help in tracking the virus’s genetic alteration (mutation) over time, which will determine the value of any vaccine in the future time (Waltz, [49]).

4. Different AI Tools

It has been documented that the National Institute of Allergy and Infectious Diseases funded the first clinical trial of an AI-based flu vaccine in 2019 in the United States (Ahuja et al., [71]). Flinders University scientists created the vaccine with the use of an AI tool called synthetic chemist, which generated trillions of synthetic chemical compounds. The researchers next used the Search Algorithm for Ligands (SAM), an AI program that sifts through trillions of molecules to determine which one might be a good vaccine adjuvant candidate (Park, [71]). This method can lessen the time it takes to develop a vaccine by several years. Screening compounds as potential adjuvants for the SARS-CoV-2 vaccine as well as a screening of new compounds based on modeling of probable changes or mutations to the novel coronavirus is easier with an AI-based approach. As the virus is having the potential to mutate, this will aid in the development of vaccines (Ahuja et al., [71]). It has never before been seen in human history for such a race to develop a vaccine against a virus. But by utilizing the power of AI, the velocity of discovery can be greatly enhanced. Figure 4 shows the process of discovering vaccine candidates via the AI/ML method.

Over the last two decades, machine learning has also aided vaccine development. Machine learning-provided ligand-protein interaction, reaction prediction (Fooshee et al., [72]), activity prediction (Zhavoronkov et al., [73]), and compound property prediction (Ma et al., [74]) are the most affected domains of vaccine and drug discovery (Chen et al., [75]). RV is a technique for developing novel vaccines that begin with pathogen genome sequencing. Through bioinformatics analysis of the pathogen genome, RV tries to identify potential vaccine candidates. By selecting epitopes and screening vaccine candidates in silico, RV can be used to choose an antigen for a novel vaccine that can elicit an immunological response and also speed up and lower the cost of the process of vaccine development (Rappuoli et al., [22]; Mora et al., [76]; Hwang et al., [77]).

VaxiJen was the first application of machine learning in RV methods, and it has shown encouraging antigen prediction outcomes (Doytchinova and Flower, [78]; Heinson et al., [79]). Vaxign, the first web-based RV program (He et al., [80]), has been used to predict vaccine candidates against a variety of viral and bacterial infections (Xiang and He, [81]). Vaxign’s first generation uses a filtering-based strategy to choose vaccine antigen candidates based on the user’s past knowledge of the pathophysiology of the target pathogen (Ong et al., [21]). Recently, Vaxign-ML, a machine learning approach, has been developed to enhance prediction accuracy (Ong et al., [82]).Vaxign-ML used the biological and physicochemical properties of protein sequences as input variables to train five different machine learning models. The input protein sequences were taken from the Protegen database, which has been collecting and annotating experimentally confirmed protective antigens for the past ten years (Yang et al., [83]; Ong et al., [21]). In a study, data was gathered from the Immune Epitope Database (IEDB), the Virus Pathogen Resource, and the National Center for Biotechnical Information by a team from the University of Southern California. Over 600,000 epitomes from 3,600 distinct species are stored in the IEDB. When applied to SARS-CoV-2, the AI model immediately ruled out 95% of the elements that could be COVID therapies, highlighting the best alternatives. This AI tool predicted a total of 26 possible potential vaccines to combat the deadly infections. The researchers chose 11 of the 26 to use in the development of a multiepitope vaccine that targets the viral spike proteins that are important for replication. The suggested vaccine design framework can address the three most commonly observed mutations and may be expanded to include other potentially unknown mutations. Several vaccines have been in use now, but in case the mutation occurs and possibly reduces the effectiveness of vaccines in use, the AI-assisted method will be able to provide quick results to design other preventive mechanisms. Currently, AI technology only employs B cell and T cell epitopes to generate findings. IgPred is a tool that predicts when immunoglobulin subclass a B cell epitope is capable of. The tool was trained using the support vector machines (SVM) method on over 14,000 epitopes and can be used to detect epitopes that induce IgG and IgA antibodies (Gupta et al., [84]; Khairkhah et al., [85]). NetCTLpan has been used in multiple SARS-CoV-2 vaccine development studies, and it provides end-to-end cytotoxic T cell epitope predictions (Mishra, [86]; Ayyagari et al., [87]). AI technology can create a stronger and faster vaccine if given more datasets and viable combinations. This approach is thought to be capable of accurately predicting over 700,000 distinct proteins (Komarraju, [88]). It is critical to forecast the peptides that bind multiple human leukocyte antigen (HLA) molecules to build and develop an effective vaccine for a large population (Brusic et al., [89]). MHC-I and MHC-II proteins encoded by the HLA gene present epitopes as antigenic determinants. Recursive feature elimination (RFE), SVM, and random forest (RF) are examples of machine learning algorithms that have been used to identify antigens from protein sequences (Bowick et al., [90]; Rahman et al., [91]). Detecting the presence of antigenic peptides presented by MHC-II is one of the most straightforward applications of ML and other AI-based technologies in vaccine development. The examples which can predict antigen presentation are MoDec and MARIA (major histocompatibility complex analysis with recurrent integrated architecture). To better understand natural immunity, various AI-related technologies have been employed to examine SARS-CoV-2 viral peptide presentation on MHC molecules from patients. Thus, it could aid either directly or indirectly in the discovery of COVID-19’s unique immune response and the development of a successful vaccine. MARIA improves HLA-II prediction by integrating better training data with a novel supervised machine learning model that uses a multimodal recurrent neural network. (RNN). A similar technique to the convolutional neural network (CNN) is a motif deconvolution technique known as MoDec. It has been used to find out peptide cleavage and MHC-II binding motifs from MS-based peptidome datasets that compromises HLA-DP, HLA-DQ, and HLA-DR alleles. To detect B cell and T cell epitopes of SARS-CoV-2, Fast et al. [92] used two artificial neural network techniques known as MARIA and NetMHCPan4. The technique discovered 405 T cell epitopes with high MHC-I and MHC-II presentation scores, as well as two putative neutralizing B cell epitopes on the S protein. This discovery will aid in the development of effective COVID-19 vaccines and neutralizing antibodies (Racle et al., [93]; Chen et al., [94]; Moore et al., [95]; Arora et al., [96]). A recent study has used a combination of computational techniques and immunoinformatics, thus identifying antigenic epitopes in the structural proteins of SARS-CoV-2 (S, E, M) and suggested a plausible multiepitope-based subunit vaccine (MESV). MESV has sufficient structural and physicochemical characteristics to activate all components of the host immune system. It also seems to have a very stable interaction with the innate immune receptor toll-like receptor-3, making it more likely to enter the host immune system. By the use of computational tools, this study could help researchers save time and money when studying experimental epitope targets (Tahir ulQamar et al., [54]). The development of AI algorithms for determining whether a peptide binds numerous HLA molecules is extremely important in making the design of vaccines more time-efficient. The proposed systems include systems based on hidden Markov models (HMMs), artificial neural networks (ANNs) (Brusic et al., [97]), and SVMs (Bozic et al., [98]). SVM has been used to predict antigens in an RV problem (Heinson et al., [79]). For both MHC-I AND MHC-II binding peptides, RANKPEP provides a position-specific score matrix (Reche et al., [99]). MHCnuggets is a neural network model based on MHC binders that have been trained on common and rare alleles (Shao et al., [100]). Lopez-Rincon et al. [101] proposed a CNN to classify 553 genome sequences with promising accuracy results for analyzing COVID-19 gene sequences. By giving a wide range of T cell epitopes, Monte Carlo-based simulation has been applied to forecast blueprinting for SARS-CoV-2 vaccines (Malone et al., [102]). It has been documented that the neural network method, NetMHC, predicts which peptides will bind and hence identify epitopes for the SARS-CoV-2 vaccine (Prachar et al., [103]). The application of AI in the research of vaccines and COVID-19 treatment is gaining a lot of attention due to international projects such as CoronaDB-AI, a data collection with genomic features that can be used to train AI models for COVID-19 treatments (KeshavarziArshadi et al., [104]; Liu et al., [105]). Figure 5 shows AI-based vaccine development for COVID-19. Recent studies have used hidden Markov models, Monte Carlo-based simulation, and neural network approaches to predict epitopes, the portion of an antigen that might stimulate an immune response, as a potential target in the development of a vaccine (Crooke et al., [106]; Prachar et al., [103]; Malone et al., [102]). A deep learning approach (DeepVacPred) has been used for predicting and designing a multiepitope vaccine that could predict 26 different SARS-CoV-2-spike-protein sequence vaccine components (Yang et al., [83]). Deep convolutional neural networks have proven to be a more reliable alternative for predicting MHC and peptide binding (Han and Kim, [107]). Deep learning autoencoders have shown promise in extracting features of human Leukocyte Antigen (HLA-A), which could be used in the development of a vaccine (Miyake et al., [108]). The network of long short-term memory has also shown some encouraging results. This type of RNN was used to predict epitopes for spikes (Abbasi et al., [109]). Malone et al. [102] used a similar strategy, employing deep learning RNN and simulated spike sequences to discover potential vaccination targets. RNN gave sequences for a protein of interest with a high degree of sequence identity. Malone et al. [102] used BepiPred, IEDB, and NEC Immune Profiler tools to create an epitope map for different HLA alleles and studied the complete SARS-CoV-2 proteome beyond spike, to give a comprehensive vaccine design blueprint for SARS-CoV-2. For the development of a vaccine for COVID-19, NLP models, notably language modeling techniques, have made a great impact. Pretrained transformers were utilized in carbohydrate chemistry to predict protein interaction (Nambiar et al., [110]) and model molecular processes (Abbasi et al., [109]), which can be applied in the vaccine development process. The transformers were employed to repurpose commercially available medications by anticipating their interactions with SARS-CoV-2 viral proteins (Beck et al., [111]). Researchers at the University of Basel in Switzerland utilized a protein-modeling tool called the Swiss Model to anticipate the architectures of proteins on the outer surface of the SARS-CoV-2 virus when the pandemic struck. DeepMind, a London-based AI firm, used its AlphaFold neural network to predict the three-dimensional form of SARS-CoV-2 proteins based on the virus’s genomic code (Waltz, [49]). Their predictions turned out to be accurate when compared to the virus’s actual protein structures. Table 2 shows some of the common tools for the prediction of epitopes and their evaluated accuracy or reported accuracy, if available.

Therefore, the AI-powered technology offered a ray of hope by assisting scientists and medical health providers in dealing with this deadly disease. To summarize the role of AI in research, AI helps in gathering and synthesizing the information, determining the cause of the disease, and selecting and developing potential drug/vaccine candidates. Furthermore, after the selection of a potential candidate, AI helps in clinical trials of the vaccine.

5. AI in Clinical Trials of the Vaccine

The most time-consuming aspect of vaccine research and development is the testing of a vaccine. To acquire a better knowledge of the disease and for the research of a potential vaccine candidate and accelerate its speed, computational approaches and AI techniques have been developed as shown in Figure 6. Several applications of AI help in the rapid classification of novel viruses by detecting their intrinsic genomic structures and can be used to recognize similarities with other pathogens (Randhawa et al., [117]). Only one out of ten molecules get approved for entering into the clinical trials, which is a huge loss for the industry (Hay et al., [118]). The reason for these failures can be caused by a lack of technical needs or infrastructure and poor patient selection. However, with a large amount of digital medical data available, AI can help in reducing these failures (Harrer et al., [119]). Moreover, different AI algorithms help in identifying the potential candidate for clinical trials of the vaccine in an efficient time after the screening of several candidates. Some researchers applied SDQ for reviewing the data of clinical trials in less than 24 hours for the same type of evaluation instead of more than 30 days. The rapid data access was accompanied by exceptional levels of good data quality resulting in a dependable outcome. As a result, the time between molecules to market was reduced from ten years to one year (Kulkarni et al., [120]). After the research for a novel vaccine candidate, it has to undergo a development process. The process is divided into preclinical and clinical trials by regulatory bodies all around the world (Singh and Mehta, [121]). The World Health Organization (WHO), the US Food and Drug Administration (USFDA), and the European Medicines Agency (EMA) have issued guidelines to plan the clinical development path of a potential vaccine candidate. Each vaccine develops in its way, based on the factors including the type of vaccine (peptide/DNA/RNA/inactivated/live), target population, and disease epidemiology. A vaccine candidate normally undergoes three phases of human development known as clinical trials, which are Phase I, Phase II, and Phase III trials before regulatory approval. To monitor the safety and efficacy in the population Phase IV trial is used after the completion of Phase III trials (Farrington and Miller, [122]; WHO technical report, [123]; Hudgens et al., [124]; Collins, [125]).

5.1. Different Phases during Clinical Trials of Vaccine

5.1.1. Phase I (20-80 Subjects)

The phase I trial involves healthy subjects. It consists of the administration of the vaccine to subjects. The aim of this trial is the detection of safety and collection of an immune response. The immunization schedule, model of vaccine administered, and dose are often evaluated (Who technical report, [123]; Hudgens et al., [124]; EMEA, 2005).

5.1.2. Phase II (100-300 Subjects)

After reaching a successful outcome in the phase I trial in terms of both immunogenicity and safety, a vaccine candidate should proceed to phase II clinical evaluation. The goal is to provide more information on safety, immunogenicity, and data on the optimal dose, vaccine preparation, and schedule of the vaccine to be taken up for the phase III confirmatory trial (Artaud et al., [126]).

5.1.3. Phase III (Large-Scale Population, 300-3000 Subjects)

The effect of a final formulation is assessed in the phase III trial and is very important for the registration and approval of a vaccine to market. Safety and efficacy are the main goals of these trials. If the vaccine’s efficacy and safety are demonstrated in the phase III trial, the manufacturer of the vaccine can submit an application to the national regulatory authority for a license and commercialize the product (Singh and Mehta, [121]).

5.1.4. Phase IV (Several Thousand People)

It is also known as postmarketing surveillance studies (PMS). This trial is used to continue to monitor the vaccine for safety and effectiveness in the population. It is carried out after the successful completion of phase III trials and following licensure of the product (Singh and Mehta, [121]).

The various vaccine candidates are classified according to the technology used for the vaccine development. The list of a few vaccines with their characteristics, manufacturer name, phase 3 trial data, and efficacy data are presented in Table 3 (Kyriakidis et al., [127]).

The majority of the work turns to testing once a vaccine candidate has been designed and developed. Vaccines are initially evaluated in the lab on cells and animals known as preclinical trials, then on human beings known as clinical trials. Unfortunately, AI tools cannot take the place of those time-consuming procedures. However, by applying patient-specific genome-exposome profile analysis, AI can help choose only a certain diseased population for engagement in Phase II and III clinical trials, allowing for early prediction of the viable therapeutic targets in the chosen patients (Mak and Pichika, [128]; Harrer et al., [119]). Predicting lead compounds and preclinical discovery of molecules before the start of clinical trials using AI tools such as predictive ML aid in the early prediction of lead molecules that will pass clinical trials with a selected population of the patient (Harrer et al., [119]). AiCure helps in monitoring the regular medication intake by patients with schizophrenia in phase II trials, resulting in a 25% increase in patient adherence and ensuring the clinical trial’s success (Mak and Pichika, [128]). NLP can help in extracting and analyzing important data from patients’ EHR records, can compare it to eligibility criteria for ongoing trials, and suggests matching studies. AI tools might be able to forecast which antigens the immune system will encounter, but what the immune system will do in a live human is beyond today’s computer capabilities. Because the human body is so complex, AI tools cannot predict what the vaccine candidate will do for the body with reliable data. However, there was no evidence found that clinical trials were conducted using computational supervision. Although AI cannot anticipate the outcome of clinical trials, it can make sense of the mountains of data generated by these trials by analyzing all the factors and identifying patterns that a human brain might miss. As thousands of patients will be engaged as the vaccine candidate progress to the second and third phases of clinical testing, AI tools or systems will be very much critical in quickly assessing clinical and immunological data (Waltz, [49]; Piccialli et al., [129]). To summarize the role of AI in clinical development, it helps in trial planning, optimizing the recruitment process, risk monitoring, toxicity prediction, and also monitoring of drug adherence.

6. Conclusion

This paper summarized the application of AI/ML in the field of research and development of the SARS-COV-2 vaccines. AI has been shown as an emerging and promising technology for detecting early coronavirus infection and monitoring the state of affected individuals. AI techniques including ML and DL have shown to be beneficial to aid in the study of the virus by examining the data available and assisting in the development of proper treatment regimens as well as in the development of a novel vaccine candidate. The AI algorithms have become more important for advanced analysis and translation of basic discoveries into novel vaccine candidates due to the large accessible amount of data. However, AI approaches cannot replace the time taking tasks such as laboratory experimentations and clinical trials, but they can aid in the planning of a trial, monitoring, and predictions of deleterious and risk factors. This paper suggested the various computational tools and techniques developed based on AI and ML which can anticipate complicated immune system activities, such as B cell and T cell epitope prediction. With the help of computer-based tools and algorithms, it considerably improves decision-making, and treatment uniformity, and resulted in the speedup in vaccine development and research to fight against the SARS-COV-2 virus.

Data Availability

No data were used.

Conflicts of Interest

The authors declare no conflict of interest.

References

M. R. Hilleman, “Vaccines in historic evolution and perspective: a narrative of vaccine discoveries,” Journal of Human Virology, vol. 3, no. 2, pp. 63–76, 2000.
View at: Publisher Site | Google Scholar
E. D. Kilbourne, “Influenza pandemics of the 20th century,” Emerging Infectious Diseases, vol. 12, no. 1, pp. 9–14, 2006.
View at: Publisher Site | Google Scholar
L. O. Gostin, D. Lucey, and A. Phelan, “The Ebola Epidemic,” JAMA, vol. 312, no. 11, pp. 1095-1096, 2014.
View at: Publisher Site | Google Scholar
A. R. Fehr and S. Perlman, “Coronaviruses: an overview of their replication and pathogenesis,” Methods in Molecular Biology, vol. 1282, pp. 1–23, 2015.
View at: Publisher Site | Google Scholar
J. S. Kahn and K. McIntosh, “History and Recent Advances in Coronavirus Discovery,” The Pediatric Infectious Disease Journal, vol. 24, Supplement 11, pp. S223–S227, 2005.
View at: Publisher Site | Google Scholar
R. J. Khan, R. K. Jha, G. M. Amera et al., “Targeting SARS-CoV-2: a systematic drug repurposing approach to identify promising inhibitors against 3C-like proteinase and -O-ribose methyltransferase,” J Biomolecular Structure & Dynamics, vol. 39, no. 8, pp. 2679–2692, 2021.
View at: Publisher Site | Google Scholar
R. Hilgenfeld and M. Peiris, “From SARS to MERS: 10 years of research on highly pathogenic human coronaviruses,” Antiviral Research, vol. 100, no. 1, pp. 286–295, 2013.
View at: Publisher Site | Google Scholar
W. H. Kong, Y. Li, M. W. Peng et al., “SARS-CoV-2 detection in patients with influenza-like illness,” Nature Microbiology, vol. 5, no. 5, pp. 675–678, 2020.
View at: Publisher Site | Google Scholar
Worldometer, “Coronavirus Cases,” 2020, https://www.worldometers.info/coronavirus/coronavirus-cases/#daily-cases.
View at: Google Scholar
Z. Jin, X. Du, Y. Xu et al., “Structure of M^pro from SARS-CoV-2 and discovery of its inhibitors,” Nature, vol. 582, no. 7811, pp. 289–293, 2020.
View at: Publisher Site | Google Scholar
M. Kandeel and M. Al-Nazawi, “Virtual screening and repurposing of FDA approved drugs against COVID-19 main protease,” Life Sciences, vol. 251, article 117627, 2020.
View at: Publisher Site | Google Scholar
B. K. Shoichet, “Virtual screening of chemical libraries,” Nature, vol. 432, no. 7019, pp. 862–865, 2004.
View at: Publisher Site | Google Scholar
E. Lionta, G. Spyrou, D. K. Vassilatis, and Z. Cournia, “Structure-based virtual screening for drug discovery: principles, applications and recent advances,” Current Topics In Medicinal Chemistry, vol. 14, no. 16, pp. 1923–1938, 2014.
View at: Publisher Site | Google Scholar
W. Yu and A. D. MacKerell, “Computer-aided drug design methods,” In Antibiotics, Humana Press, New York, NY, vol. 1520, pp. 85–106, 2017.
View at: Publisher Site | Google Scholar
A. KeshavarziArshadi, M. Salem, J. Collins, J. S. Yuan, and D. Chakrabarti, “DeepMalaria: artificial intelligence driven discovery of potent antiplasmodials,” Frontiers in Pharmacology, vol. 10, p. 1526, 2020.
View at: Publisher Site | Google Scholar
A. W. Senior, R. Evans, J. Jumper et al., “Improved protein structure prediction using potentials from deep learning,” Nature, vol. 577, no. 7792, pp. 706–710, 2020.
View at: Publisher Site | Google Scholar
H. Zhang, K. M. Saravanan, Y. Yang et al., “Deep learning based drug screening for novel coronavirus 2019-nCov,” Interdisciplinary sciences, computational life sciences, vol. 12, no. 3, pp. 368–376, 2020.
View at: Publisher Site | Google Scholar
F. Li, “Structure, function, and evolution of coronavirus spike proteins,” Annual review of virology, vol. 3, no. 1, pp. 237–261, 2016.
View at: Publisher Site | Google Scholar
R. L. Roper and K. E. Rehm, “SARS vaccines: where are we?” Expert Review of Vaccines, vol. 8, no. 7, pp. 887–898, 2009.
View at: Publisher Site | Google Scholar
E. de Wit, N. van Doremalen, D. Falzarano, and V. J. Munster, “SARS and MERS: recent insights into emerging coronaviruses,” Nature Reviews. Microbiology, vol. 14, no. 8, pp. 523–534, 2016.
View at: Publisher Site | Google Scholar
E. Ong, M. F. Cooke, A. Huffman et al., “Vaxign2: the second generation of the first Web-based vaccine design program using reverse vaccinology and machine learning,” Nucleic Acids Research, vol. 49, no. W1, pp. W671–W678, 2021.
View at: Publisher Site | Google Scholar
R. Rappuoli, “Reverse vaccinology,” Current Opinion in Microbiology, vol. 3, no. 5, pp. 445–450, 2000.
View at: Publisher Site | Google Scholar
J. Bullock, L. Alexandra, K. H. Pham, C. S. N. Lam, and M. Luengo-Oroz, “Mapping the landscape of artificial intelligence applications against COVID-19,” 2020, https://arxiv.org/abs/2003.11336.
View at: Google Scholar
R. E. Soria-Guerra, R. Nieto-Gomez, D. O. Govea-Alonso, and S. Rosales-Mendoza, “An overview of bioinformatics tools for epitope prediction: implications on vaccine development,” Journal of Biomedical Informatics, vol. 53, pp. 405–414, 2015.
View at: Publisher Site | Google Scholar
A. I. Heinson, C. H. Woelk, and M. L. Newell, “The promise of reverse vaccinology,” International Health, vol. 7, no. 2, pp. 85–89, 2015.
View at: Publisher Site | Google Scholar
L. Bruno, M. Cortese, R. Rappuoli, and M. Merola, “Lessons from reverse vaccinology for viral vaccine design,” Current Opinion in Virology, vol. 11, pp. 89–97, 2015.
View at: Publisher Site | Google Scholar
T. Folaranmi, L. Rubin, S. W. Martin, M. Patel, J. R. MacNeil, and Centers for Disease Control (CDC), “Use of serogroup B meningococcal vaccines in persons aged ≥10 years at increased risk for serogroup B meningococcal disease: recommendations of the advisory committee on immunization practices, 2015,” MMWR. Morbidity and Mortality Weekly Report, vol. 64, no. 22, pp. 608–612, 2015.
View at: Google Scholar
J. McCarthy, “Artificial intelligence, logic and formalizing common sense,” Philosophical Logic and Artificial Intelligence, Springer, Dordrecht, 1989.
View at: Publisher Site | Google Scholar
A. M. Turing, “Computing Machinery and Intelligence,” In: Parsing the Turing Test, Springer, pp. 23–65, 2009.
View at: Publisher Site | Google Scholar
S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, Pearson Education, Inc, 3rd edition, 2010.
Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436–444, 2015.
View at: Publisher Site | Google Scholar
D. Grover and B. Toghi, “MNIST dataset classification utilizing k-NN classifier with modified sliding-window metric,” Advances in Computer Vision CVC, Springer, Cham, p. 944, 2020.
View at: Publisher Site | Google Scholar
Y. Sun, D. Liang, X. Wang, and X. Tang, “DeepID3: face recognition with very deep neural networks,” 2015, https://arxiv.org/abs/1502.00873.
View at: Google Scholar
K. Theofilatos, N. Pavlopoulou, C. Papasavvas et al., “Predicting protein complexes from weighted protein-protein interaction graphs with a novel unsupervised methodology: evolutionary enhanced Markov clustering,” Artificial Intelligence in Medicine, vol. 63, no. 3, pp. 181–189, 2015.
View at: Publisher Site | Google Scholar
T. Rapakoulia, K. Theofilatos, D. Kleftogiannis, S. Likothanasis, A. Tsakalidis, and S. Mavroudi, “EnsembleGASVR: a novel ensemble method for classifying missense single nucleotide polymorphisms,” Bioinformatics, vol. 30, no. 16, pp. 2324–2333, 2014.
View at: Publisher Site | Google Scholar
D. Duvenaud, D. Maclaurin, J. Aguilera-Iparraguirre et al., “Convolutional networks on graphs for learning molecular fingerprints. arXiv,” 2015, https://arxiv.org/abs/1509.09292.
View at: Google Scholar
S. Kearnes, K. McCloskey, M. Berndl, V. Pande, and P. Riley, “Molecular graph convolutions: moving beyond fingerprints,” Journal of Computer-Aided Molecular Design, vol. 30, no. 8, pp. 595–608, 2016.
View at: Publisher Site | Google Scholar
A. Fout, J. Byrd, B. Shariat, and A. Ben-Hur, “Protein interface prediction using graph convolutional networks,” Advances in Neural Information Processing Systems (Long Beach, CA), vol. 30, pp. 6530–6539, 2017.
View at: Google Scholar
K. Liu, X. Sun, L. Jia et al., “Chemi-net: A molecular graph convolutional network for accurate drug property prediction,” International Journal of Molecular Sciences, vol. 20, no. 14, p. 3389, 2019.
View at: Publisher Site | Google Scholar
J. Devlin, W. M. Chang, K. Lee, K. T. Google, and A. I. Language, “BERT: pre-training of deep bidirectional transformers for language understanding. arXiv,” 2018, https://arxiv.org/abs/1810.04805.
View at: Google Scholar
B. W. Neuman, G. Kiss, A. H. Kunding et al., “A structural analysis of M protein in coronavirus assembly and morphology,” Journal of Structural Biology, vol. 174, no. 1, pp. 11–22, 2011.
View at: Publisher Site | Google Scholar
T. Tang, M. Bidon, J. A. Jaimes, G. R. Whittaker, and S. Daniel, “Coronavirus membrane fusion mechanism offers a potential target for antiviral development,” Antiviral Research, vol. 178, article 104792, 2020.
View at: Publisher Site | Google Scholar
P. Venkatagopalan, S. M. Daskalova, L. A. Lopez, K. A. Dolezal, and B. G. Hogue, “Coronavirus envelope (E) protein remains at the site of assembly,” Virology, vol. 478, pp. 75–85, 2015.
View at: Publisher Site | Google Scholar
J. L. Nieto-Torres, M. L. Dediego, E. Alvarez et al., “Subcellular location and topology of severe acute respiratory syndrome coronavirus envelope protein,” Virology, vol. 415, no. 2, pp. 69–82, 2011.
View at: Publisher Site | Google Scholar
C. A. deHaan and P. J. Rottier, “Molecular interactions in the assembly of coronaviruses,” Advances in Virus Research, vol. 64, pp. 165–230, 2005.
View at: Publisher Site | Google Scholar
J. Tooze, S. Tooze, and G. Warren, “Replication of coronavirus MHV-A59 in sac- cells: determination of the first site of budding of progeny virions,” European Journal of Cell Biology, vol. 33, no. 2, pp. 281–293, 1984.
View at: Google Scholar
R. J. de Groot, “Structure, function and evolution of the hemagglutinin-esterase proteins of corona- and toroviruses,” Glycoconjugate Journal, vol. 23, no. 1-2, pp. 59–72, 2006.
View at: Publisher Site | Google Scholar
Q. Zeng, M. A. Langereis, A. L. van Vliet, E. G. Huizinga, and R. J. de Groot, “Structure of coronavirus hemagglutinin-esterase offers insight into corona and influenza virus evolution,” Proceedings of the National Academy of Sciences of the United States of America, vol. 105, no. 26, pp. 9065–9069, 2008.
View at: Publisher Site | Google Scholar
E. Waltz, “AI takes its best shot: what AI can—and can't—do in the race for a coronavirus vaccine - [vaccine],” IEEE Spectrum, vol. 57, no. 10, pp. 24–67, 2020.
View at: Publisher Site | Google Scholar
S. Saha and G. P. Raghava, “Prediction of continuous B-cell epitopes in an antigen using recurrent neural network,” Proteins, vol. 65, no. 1, pp. 40–48, 2006.
View at: Publisher Site | Google Scholar
J. Ponomarenko, H. H. Bui, W. Li et al., “ElliPro: a new structure-based tool for the prediction of antibody epitopes,” BMC Bioinformatics, vol. 9, no. 1, p. 514, 2008.
View at: Publisher Site | Google Scholar
M. O. Rafi, K. Al-Khafaji, M. T. Sarker, T. Taskin-Tok, A. S. Rana, and M. S. Rahman, “Design of a multi-epitope vaccine against SARS-CoV-2: immunoinformatic and computational methods,” RSC Advances, vol. 12, no. 7, pp. 4288–4310, 2022.
View at: Publisher Site | Google Scholar
M. Zhang, K. Ishii, H. Hisaeda et al., “Ubiquitin-fusion degradation pathway plays an indispensable role in naked DNA vaccination with a chimeric gene encoding a syngeneic cytotoxic T lymphocyte epitope of melanocyte and green fluorescent protein,” Immunology, vol. 112, no. 4, pp. 567–574, 2004.
View at: Publisher Site | Google Scholar
M. Tahir UlQamar, F. Shahid, S. Aslam et al., “Reverse vaccinology assisted designing of multiepitope-based subunit vaccine against SARS-CoV-2,” Infectious Diseases of Poverty, vol. 9, no. 1, p. 132, 2020.
View at: Publisher Site | Google Scholar
C. J. Melief, T. van Hall, R. Arens, F. Ossendorp, and S. H. van der Burg, “Therapeutic cancer vaccines,” The Journal of Clinical Investigation, vol. 125, no. 9, pp. 3401–3412, 2015.
View at: Publisher Site | Google Scholar
K. F. Bol, E. H. Aarntzen, J. M. Pots et al., “Prophylactic vaccines are potent activators of monocyte-derived dendritic cells and drive effective anti-tumor responses in melanoma patients at the cost of toxicity,” Cancer Immunology, Immunotherapy: Cii, vol. 65, no. 3, pp. 327–339, 2016.
View at: Publisher Site | Google Scholar
J. Brotherton, “HPV prophylactic vaccines: lessons learned from 10 years experience,” Future Virology, vol. 10, no. 8, pp. 999–1009, 2015.
View at: Publisher Site | Google Scholar
I. H. Frazer, “Development and implementation of papillomavirus prophylactic vaccines,” Journal of Immunology, vol. 192, no. 9, pp. 4007–4011, 2014.
View at: Publisher Site | Google Scholar
F. Amanat and F. Krammer, “SARS-CoV-2 vaccines: status report,” Immunity, vol. 52, no. 4, pp. 583–589, 2020.
View at: Publisher Site | Google Scholar
B. Deb, H. Shah, and S. Goel, “Current global vaccine and drug efforts against COVID-19: pros and cons of bypassing animal trials,” Journal of Biosciences, vol. 45, no. 1, p. 82, 2020.
View at: Publisher Site | Google Scholar
T. Thanh Le, Z. Andreadakis, A. Kumar et al., “The COVID-19 vaccine development landscape,” Nature Reviews Drug Discovery, vol. 19, no. 5, pp. 305-306, 2020.
View at: Publisher Site | Google Scholar
J. Cohen, “Vaccine designers take first shots at COVID-19,” Science, vol. 368, no. 6486, pp. 14–16, 2020.
View at: Publisher Site | Google Scholar
A. Gast, “COVID showed AI's potential for accelerating progress, but we must ensure AI is employed in ways that are trusted,” 2022, https://theprint.in/tech/ai-helped-moderna-speed-up-covid-vaccine-development-now-it-can-help-climate-too/955061/.
View at: Google Scholar
A. Beatrice, “Artificial Intelligence, A major factor behind Pfizer’s US$900M profit,” 2021, https://www.analyticsinsight.net/artificial-intelligence-a-major-factor-behind-pfizers-us900m-profit/.
View at: Google Scholar
R. Kumar and K. Veer, “How artificial intelligence and internet of things can aid in the distribution of COVID-19 vaccines,” Diabetes & metabolic syndrome, vol. 15, no. 3, pp. 1049-1050, 2021.
View at: Publisher Site | Google Scholar
A. Beatrice, “From Vaccine to drug making, Astrazeneca is relying on AI for growth,” 2021, https://www.analyticsinsight.net/from-vaccine-to-drug-making-astrazeneca-is-relying-on-ai-for-growth/#:~:text=AstraZeneca%2C%20one%20of%20the%20biggest,also%20in%20other%20drug%20discoveries.
View at: Google Scholar
N. Lurie, M. Saville, R. Hatchett, and J. Halton, “Developing COVID-19 vaccines at pandemic speed,” The New England Journal of Medicine, vol. 382, no. 21, pp. 1969–1973, 2020.
View at: Publisher Site | Google Scholar
J. Weatherall, “Data Science & Artificial Intelligence: unlocking new science insights,” 2021, https://www.astrazeneca.com/r-d/data-science-and-ai.html.
View at: Google Scholar
S. Sachdeva, “How can India devise a seamless vaccination program using AI & IoT?” 2021, https://www.geospatialworld.net/blogs/how-can-india-devise-a-seamless-vaccination-program-using-ai-iot/.
View at: Google Scholar
O. Peckham, “Pfizer Discusses Use of Supercomputing and AI for Covid Drug Development,” 2022, https://www.hpcwire.com/2022/03/24/pfizer-discusses-use-of-supercomputing-and-ai-for-covid-drug-development/.
View at: Google Scholar
A. S. Ahuja, V. P. Reddy, and O. Marques, “Artificial intelligence and COVID-19: a multidisciplinary approach,” Integrative Medicine Research, vol. 9, no. 3, article 100434, 2020.
View at: Publisher Site | Google Scholar
D. Fooshee, A. Mood, E. Gutman et al., “Deep learning for chemical reaction prediction,” Molecular Systems Design & Engineering, vol. 3, no. 3, pp. 442–452, 2018.
View at: Publisher Site | Google Scholar
A. Zhavoronkov, Y. A. Ivanenkov, A. Aliper et al., “Deep learning enables rapid identification of potent DDR1 kinase inhibitors,” Nature Biotechnology, vol. 37, no. 9, pp. 1038–1040, 2019.
View at: Publisher Site | Google Scholar
J. Ma, R. P. Sheridan, A. Liaw, G. E. Dahl, and V. Svetnik, “Deep neural nets as a method for quantitative structure-activity relationships,” Journal of Chemical Information and Modeling, vol. 55, no. 2, pp. 263–274, 2015.
View at: Publisher Site | Google Scholar
H. Chen, O. Engkvist, Y. Wang, M. Olivecrona, and T. Blaschke, “The rise of deep learning in drug discovery,” Drug Discovery Today, vol. 23, no. 6, pp. 1241–1250, 2018.
View at: Publisher Site | Google Scholar
M. Mora, D. Veggi, L. Santini, M. Pizza, and R. Rappuoli, “Reverse vaccinology,” Drug Discovery Today, vol. 8, no. 10, pp. 459–464, 2003.
View at: Publisher Site | Google Scholar
W. Hwang, W. Lei, N. M. Katritsis, M. MacMahon, K. Chapman, and N. Han, “Current and prospective computational approaches and challenges for developing COVID-19 vaccines,” Advanced Drug Delivery Reviews, vol. 172, pp. 249–274, 2021.
View at: Publisher Site | Google Scholar
I. A. Doytchinova and D. R. Flower, “VaxiJen: a server for prediction of protective antigens, tumour antigens and subunit vaccines,” BMC Bioinformatics, vol. 8, no. 1, p. 4, 2007.
View at: Publisher Site | Google Scholar
A. I. Heinson, Y. Gunawardana, B. Moesker et al., “Enhancing the biological relevance of machine learning classifiers for reverse vaccinology,” International Journal of Molecular Sciences, vol. 18, no. 2, p. 312, 2017.
View at: Publisher Site | Google Scholar
Y. He, Z. Xiang, and H. L. Mobley, “Vaxign: the first web-based vaccine design program for reverse vaccinology and applications for vaccine development,” Journal of Biomedicine & Biotechnology, vol. 2010, Article ID 297505, 2010.
View at: Publisher Site | Google Scholar
Z. Xiang and Y. He, “Genome-wide prediction of vaccine targets for human herpes simplex viruses using Vaxign reverse vaccinology,” BMC Bioinformatics, vol. 14, pp. S2–S4, 2013.
View at: Publisher Site | Google Scholar
E. Ong, H. Wang, M. U. Wong, M. Seetharaman, N. Valdez, and Y. He, “Vaxign-ML: supervised machine learning reverse vaccinology model for improved prediction of bacterial protective antigens,” Bioinformatics (Oxford, England), vol. 36, no. 10, pp. 3185–3191, 2020.
View at: Publisher Site | Google Scholar
Z. Yang, P. Bogdan, and S. Nazarian, “An in silico deep learning approach to multi-epitope vaccine design: a SARS- CoV-2 case study,” Scientific Reports, vol. 11, no. 1, p. 3238, 2021.
View at: Publisher Site | Google Scholar
A. K. Gupta, M. S. Khan, S. Choudhury et al., “CoronaVR: a computational resource and analysis of epitopes and therapeutics for severe acute respiratory syndrome Coronavirus-2,” Frontiers in Microbiology, vol. 11, p. 1858, 2020.
View at: Publisher Site | Google Scholar
N. Khairkhah, M. R. Aghasadeghi, A. Namvar, and A. Bolhassani, “Design of novel multiepitope constructs-based peptide vaccine against the structural S, N and M proteins of human COVID-19 using immunoinformatics analysis,” PLoS One, vol. 15, no. 10, article e0240577, 2020.
View at: Publisher Site | Google Scholar
S. Mishra, “Designing of cytotoxic and helper T cell epitope map provides insights into the highly contagious nature of the pandemic novel coronavirus SARS-CoV-2,” Royal Society Open Science, vol. 7, no. 9, article 201141, 2020.
View at: Publisher Site | Google Scholar
V. S. Ayyagari, V. Tc, and K. Srirama, “Design of a multi-epitope-based vaccine targeting M-protein of SARS-CoV2: an immunoinformatics approach,” Journal of Biomolecular Structure and Dynamics, vol. 40, no. 7, pp. 2963–2977, 2020.
View at: Publisher Site | Google Scholar
A. Komarraju, “A new AI tool is helping researchers come up with COVID-19 vaccine,” 2021, https://www.analyticsinsight.net/a-new-ai-tool-is-helping-researchers-come-up-with-covid-19-vaccine/.
View at: Google Scholar
V. Brusic, N. Petrovsky, G. Zhang, and V. B. Bajic, “Prediction of promiscuous peptides that bind HLA class I molecules,” Immunology & Cell Biology, vol. 80, no. 3, pp. 280–285, 2002.
View at: Publisher Site | Google Scholar
G. C. Bowick and A. D. Barrett, “Comparative pathogenesis and systems biology for biodefense virus vaccine development,” Journal of Biomedicine & Biotechnology, vol. 2010, Article ID 236528, 2010.
View at: Publisher Site | Google Scholar
M. S. Rahman, M. K. Rahman, S. Saha, M. Kaykobad, and M. S. Rahman, “Antigenic: an improved prediction model of protective antigens,” Artificial Intelligence in Medicine, vol. 94, pp. 28–41, 2019.
View at: Publisher Site | Google Scholar
E. Fast, R. B. Altman, and B. Chen, “Potential T-cell and B-cell epitopes of 2019-nCoV,” Biorxiv, 2020.
View at: Publisher Site | Google Scholar
J. Racle, J. Michaux, G. A. Rockinger et al., “Robust prediction of HLA class II epitopes by deep motif deconvolution of immunopeptidomes,” Nature Biotechnology, vol. 37, no. 11, pp. 1283–1286, 2019.
View at: Publisher Site | Google Scholar
B. Chen, M. S. Khodadoust, N. Olsson et al., “Predicting HLA class II antigen presentation through integrated deep learning,” Nature Biotechnology, vol. 37, no. 11, pp. 1332–1343, 2019.
View at: Publisher Site | Google Scholar
T. V. Moore and M. I. Nishimura, “Improved MHC II epitope prediction -- a step towards personalized medicine,” Nature Reviews. Clinical Oncology, vol. 17, no. 2, pp. 71-72, 2020.
View at: Publisher Site | Google Scholar
G. Arora, J. Joshi, R. S. Mandal, N. Shrivastava, R. Virmani, and T. Sethi, “Artificial intelligence in surveillance, diagnosis, drug discovery, and vaccine development against COVID-19,” Pathogens (Basel, Switzerland), vol. 10, no. 8, p. 1048, 2021.
View at: Publisher Site | Google Scholar
V. Brusic, V. B. Bajic, and N. Petrovsky, “Computational methods for prediction of T-cell epitopes--a framework for modelling, testing, and applications,” Methods, vol. 34, no. 4, pp. 436–443, 2004.
View at: Publisher Site | Google Scholar
I. Bozic, G. Zhang, and V. Brusic, “Predictive vaccinology: optimisation of predictions using support vector machine classifiers,” Intelligent Data Engineering And Automated Learning, vol. 3578, pp. 375–381, 2005.
View at: Publisher Site | Google Scholar
P. A. Reche, J. P. Glutting, H. Zhang, and E. L. Reinherz, “Enhancement to the RANKPEP resource for the prediction of peptide binding to MHC molecules using profiles,” Immunogenetics, vol. 56, no. 6, pp. 405–419, 2004.
View at: Publisher Site | Google Scholar
X. M. Shao, R. Bhattacharya, J. Huang et al., “High-throughput prediction of MHC class I and II neoantigens with MHCnuggets,” Cancer Immunology Research, vol. 8, no. 3, pp. 396–408, 2020.
View at: Publisher Site | Google Scholar
A. Lopez-Rincon, A. Tonda, L. Mendoza-Maldonado et al., “Classification and specific primer design for accurate detection of SARS- CoV-2 using deep learning,” Scientific Reports, vol. 11, no. 1, p. 947, 2021.
View at: Publisher Site | Google Scholar
B. Malone, B. Simovski, C. Moliné et al., “Artificial intelligence predicts the immunogenic landscape of SARS-CoV-2 leading to universal blueprints for vaccine designs,” Scientific Reports, vol. 10, no. 1, p. 22375, 2020.
View at: Publisher Site | Google Scholar
M. Prachar, S. Justesen, D. B. Steen-Jensen et al., “Identification and validation of 174 COVID-19 vaccine candidate epitopes reveals low performance of common epitope prediction tools,” Scientific Reports, vol. 10, no. 1, p. 20465, 2020.
View at: Publisher Site | Google Scholar
A. KeshavarziArshadi, J. Webb, M. Salem et al., “Artificial intelligence for COVID-19 drug discovery and vaccine development,” Frontiers In Artificial Intelligence, vol. 3, p. 65, 2020.
View at: Publisher Site | Google Scholar
G. Liu, B. Carter, and D. K. Gifford, “Predicted cellular immunity population coverage gaps for SARS-CoV-2 subunit vaccines and their augmentation by compact peptide sets,” Cell Systems, vol. 12, no. 1, pp. 102–107.e4, 2021.
View at: Publisher Site | Google Scholar
S. N. Crooke, I. G. Ovsyannikova, R. B. Kennedy, and G. A. Poland, “Immunoinformatic identification of B cell and T cell epitopes in the SARS- CoV-2 proteome,” Scientific Reports, vol. 10, no. 1, p. 14179, 2020.
View at: Publisher Site | Google Scholar
Y. Han and D. Kim, “Deep convolutional neural networks for pan-specific peptide-MHC class I binding prediction,” BMC Bioinformatics, vol. 18, no. 1, p. 585, 2017.
View at: Publisher Site | Google Scholar
J. Miyake, Y. Kaneshita, S. Asatani, S. Tagawa, H. Niioka, and T. Hirano, “Graphical classification of DNA sequences of HLA alleles by deep learning,” Human Cell, vol. 31, no. 2, pp. 102–105, 2018.
View at: Publisher Site | Google Scholar
B. A. Abbasi, D. Saraf, T. Sharma et al., “Identification of vaccine targets & design of vaccine against SARS-CoV-2 coronavirus using computational and deep learning-based approaches,” PeerJ, vol. 10, article e13380, 2022.
View at: Publisher Site | Google Scholar
A. Nambiar, M. E. Heflin, S. Liu, S. Maslov, M. Hopkins, and A. Ritz, “Transforming the language of life: transformer neural networks for protein prediction tasks,” In Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, pp. 1–8, 2020.
View at: Publisher Site | Google Scholar
B. R. Beck, B. Shin, Y. Choi, S. Park, and K. Kang, “Predicting commercially available antiviral drugs that may act on the novel coronavirus (SARS-CoV-2) through a drug-target interaction deep learning model,” Computational and Structural Biotechnology Journal, vol. 18, pp. 784–790, 2020.
View at: Publisher Site | Google Scholar
J. He, F. Huang, J. Zhang et al., “Vaccine design based on 16 epitopes of SARS-CoV-2 spike protein,” Journal of Medical Virology, vol. 93, no. 4, pp. 2115–2131, 2021.
View at: Publisher Site | Google Scholar
Z. Yazdani, A. Rafiei, M. Yazdani, and R. Valadan, “Design an efficient multi-epitope peptide vaccine candidate against SARS-CoV-2: an in silico analysis,” Infection and Drug Resistance, vol. 13, pp. 3007–3022, 2020.
View at: Publisher Site | Google Scholar
K. M. Campbell, G. Steiner, D. K. Wells, A. Ribas, and A. Kalbasi, “Prioritization of SARS-CoV-2 epitopes using a pan-HLA and global population inference approach,” Bio Rxiv, 2020.
View at: Publisher Site | Google Scholar
A. Safavi, A. Kefayat, E. Mahdevar, A. Abiri, and F. Ghahremani, “Exploring the out of sight antigens of SARS-CoV-2 to design a candidate multi- epitope vaccine by utilizing immunoinformatics approaches,” Vaccine, vol. 38, no. 48, pp. 7612–7628, 2020.
View at: Publisher Site | Google Scholar
M. S. Rahman, M. N. Hoque, M. R. Islam et al., “Epitope-based chimeric peptide vaccine design against S, M and E proteins of SARS-CoV-2 etiologic agent of global pandemic COVID-19: an in silico approach,” PeerJ, vol. 8, article e9572, 2020.
View at: Publisher Site | Google Scholar
G. S. Randhawa, M. Soltysiak, H. El Roz, C. de Souza, K. A. Hill, and L. Kari, “Machine learning using intrinsic genomic signatures for rapid classification of novel pathogens: COVID-19 case study,” PLoS One, vol. 15, no. 4, article e0232391, 2020.
View at: Publisher Site | Google Scholar
M. Hay, D. W. Thomas, J. L. Craighead, C. Economides, and J. Rosenthal, “Clinical development success rates for investigational drugs,” Nature Biotechnology, vol. 32, no. 1, pp. 40–51, 2014.
View at: Publisher Site | Google Scholar
S. Harrer, P. Shah, B. Antony, and J. Hu, “Artificial intelligence for clinical trial design,” Trends In Pharmacological Science, vol. 40, no. 8, pp. 577–591, 2019.
View at: Publisher Site | Google Scholar
R. Kulkarni, P. Fersht, and D. Cushman, “Pfizer’s vaccine proves the Power of Emerging Tech when the Burning Platform is Red Hot,” 2021, https://www.hfsresearch.com/research/pfizers-vaccine-proves-the-power-of-emerging-tech-when-the-burning-platform-is-red-hot/.
View at: Google Scholar
K. Singh and S. Mehta, “The clinical development process for a novel preventive vaccine: an overview,” Journal of Postgraduate Medicine, vol. 62, no. 1, pp. 4–11, 2016.
View at: Publisher Site | Google Scholar
C. P. Farrington and E. Miller, “Vaccine trials,” Molecular Biotechnology, vol. 17, no. 1, pp. 43–58, 2001.
View at: Publisher Site | Google Scholar
WHO Technical Report, “Annex 1: WHO guidelines on clinical evaluation of vaccines: regulatory expectations,” World Health Organization, pp. 36–96, 2004, https://www.who.int/biologicals/publications/trs/areas/vaccines/clinical_evaluation/035-101.pdf.
View at: Google Scholar
M. G. Hudgens, P. B. Gilbert, and S. G. Self, “Endpoints in vaccine trials,” Statistical Methods in Medical Research, vol. 13, no. 2, pp. 89–114, 2004.
View at: Publisher Site | Google Scholar
H. Collins, “Vaccine development: from concept to licensed product. In: Kahn P, editor. AIDS Vaccine Handbook: Global Perspectives. 2nd ed. AIDS Vaccine Advocacy Coalition (AVAC), 7–43,” 2005.
View at: Google Scholar
C. Artaud, L. Kara, and O. Launay, “Vaccine development: from preclinical studies to phase 1/2 clinical trials,” Methods in Molecular Biology, vol. 2013, pp. 165–176, 2019.
View at: Publisher Site | Google Scholar
N. C. Kyriakidis, A. López-Cortés, E. V. González, A. B. Grimaldos, and E. O. Prado, “SARS-CoV-2 vaccines strategies: a comprehensive review of phase 3 candidates,” Vaccines, vol. 6, no. 1, pp. 17–28, 2021.
View at: Publisher Site | Google Scholar
K. K. Mak and M. R. Pichika, “Artificial intelligence in drug development: present status and future prospects,” Drug Discovery Today, vol. 24, no. 3, pp. 773–780, 2019.
View at: Publisher Site | Google Scholar
F. Piccialli, V. S. di Cola, F. Giampaolo, and S. Cuomo, “The role of artificial intelligence in fighting the COVID-19 pandemic. Information systems frontiers,” Information Systems Frontiers, vol. 23, no. 6, pp. 1467–1497, 2021.
View at: Publisher Site | Google Scholar

Copyright

Copyright © 2022 Ashwani Sharma et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

PDF Download Citation

Download other formats

Order printed copies

Views

1817

Downloads

1064

Citations