Ben Geoffrey

In this work, we develop a Bayesian weighted scheme to generate evolutionary lineages of a particular viral protein sequence of interest and through a process of clustering and choosing representative lineages from the different clusters according to an evolutionary fitness objective function, we demonstrate it is possible to have anticipated the emergence of the SARS-CoV 2 (2019) strain from the SARS-CoV 1(2004) strain and having shown this retrospectively, we discuss the possibility of applying this approach along with continuous genomic surveillance of SARS-CoVs to prevent or reduce severity of future SARS-CoV related pandemics by being prepared with broad neutralization strategies for anticipated future lineages of SARS-CoVs identified through bioinformatics approaches such as that reported in this work.

In this work, a Bayesian walker was constructed that generates mutations that are more prone as per UNIPROT variant data. The Bayesian walker was used to search the mutational landscape of Severe Acute Respiratory Syndrome Corona Virus 2 (SARS-CoV-2) and a computational workflow was followed to evaluate whether a particular mutation would satisfy natural selection's fitness criteria. For SARS-CoV-2, the empirical known fitness criteria derived from SARS-CoV-2 micro-evolution data is 3-fold criteria. Mutations that have emerged on the Spike protein of SARS-CoV-2 are ones that preserve the structural integrity of the Spike, retain, or increase infectivity and are Immune evasive as per literature reports. Based on the molecular mechanism of infectivity and Immune evasion of SARS-CoV-2, a molecular modelling workflow was adopted to investigate the evolutionary feasibility of the mutations generated by the walker, to check whether the mutations satisfy the 3-fold fitness criteria for SARS-CoV-2 micro-evolution. It was found that the walker (mutation generator) coupled with the computational workflow to evaluate the evolutionary fitness of the generated mutations, re-generated the mutants corresponding to the Alpha, Beta and Gamma variants of SARS-CoV-2 demonstrating the ability of the methodology to generate the micro-evolution of SARS-CoV-2. Having demonstrated the ability of the methodology to generate the micro-evolution of SARS-CoV-2, the computational methodology was used to make predictions for new mutants that could emerge.Communicated by Ramaswamy H. Sarma

The therapeutic challenge in addressing Viruses remains the fact that Viral mutations are random and often hard for our therapeutic response to keep up with in terms of coming up with drugs or vaccines for the new viral variants. Therefore, it would be a boon to our anti-viral therapy if we could somehow anticipate the viral mutations that would arise and be prepared for them. Epidemics/Pandemics are also often driven by Viruses picking up mutations that cause them to become more infectious and virulent in Humans. Therefore, it would be a great boon if we could somehow anticipate the mutation associated with a virus that would cause them to virulent in humans and be prepared therapeutically to neutralize the strains associated with these mutations. With the data associated with the evolution of SARS CoV, it is possible now to understand how viral variants emerge according to Natural selection. Natural selection picks out the fittest variants in terms of their infectivity and virulence to be the most dominant. In the case of SARS CoV 2, the emergence of new dominant variants is associated with stabilization mutations on the Spike of SARS CoV 2 that have led to increased affinity for human ACE2 receptor and decreased affinity for neutralizing antibodies. An in-silico approach has been shown that enables us to anticipate these possible dominant variants and thereby guide therapy and response in the present pandemic or can be applied more generally to prevent future pandemics.

Network data is composed of nodes and edges. Successful application of machine learning/deep learning algorithms on network data to make node classification and link prediction have been shown in the area of social networks through which highly customized suggestions are offered to social network users. Similarly one can attempt the use of machine learning/deep learning algorithms on biological network data to generate predictions of scientific usefulness. In the presented work, compound-drug target interaction network data set from bindingDB has been used to train deep learning neural network and a multi class classification has been implemented to classify PubChem compound queried by the user into class labels of PBD IDs. This way target interaction prediction for PubChem compounds is carried out using deep learning. The user is required to input the PubChem Compound ID (CID) of the compound the user wishes to gain information about its predicted biological activity and the tool outputs the RCSB PDB IDs of the predicted drug target interaction for the input CID. Further the tool also optimizes the compound of interest of the user toward drug likeness properties through a deep learning based structure optimization protocol. The tool also incorporates a feature to perform automated In Silico modelling to find the interaction between the compounds and the predicted drug targets to uncover their protein-ligand interaction profiles. The program is hosted, supported and maintained at the following GitHub repository. https://github.com/bengeof/Compound2DeNovoDrugPropMax. Anticipating the use of quantum computing and quantum machine learning in drug discovery we use the Penny-lane interface to quantum hardware to turn classical Keras layers used in our machine/deep learning models into a quantum layer and introduce quantum layers into classical models to produce a quantum-classical machine/deep learning hybrid model of our tool and the code corresponding to the same is provided below. https://github.com/bengeof/QPoweredCompound2DeNovoDrugPropMax. Communicated by Ramaswamy H. Sarma

This work seeks to combine the combined advantage of leveraging these emerging areas of Artificial Intelligence and quantum computing in applying it to solve the specific biological problem of protein structure prediction using Quantum Machine Learning algorithms. The CASP dataset from ProteinNet was downloaded which is a standardized data set for machine learning of protein structure. Its large and standardized dataset of PDB entries contains the coordinates of the backbone atoms, corresponding to the sequential chain of N, C_alpha, and C’ atoms. This dataset was used to train a quantum-classical hybrid Keras deep neural network model to predict the structure of the proteins. To visually qualify the quality of the predicted versus the actual protein structure, protein contact maps were generated with the experimental and predicted protein structure data and qualified. Therefore this model is recommended for the use of protein structure prediction using AI leveraging the power of quantum computers.

The past decade has seen a surge in the range of application data science, machine learning, deep learning, and AI methods to drug discovery. The presented work involves an assemblage of a variety of AI methods for drug discovery along with the incorporation of in silico techniques to provide a holistic tool for automated drug discovery. When drug candidates are required to be identified for aparticular drug target of interest, the user is required to provide the tool target signatures in the form of an amino acid sequence or its corresponding nucleotide sequence. The tool collects data registered on PubChem required to perform an automated QSAR and with the validated QSAR model, prediction and drug lead generation are carried out. This protocol we call Target2Drug. This is followed by a protocol we call Target2DeNovoDrug wherein novel molecules with likely activityagainst the target are generated de novo using a generative LSTM model. It is often required in drug discovery that the generated molecules possess certain properties like drug-likeness, and therefore to optimize the generated de novo molecules toward the required drug-like property we use a deep learning model called DeepFMPO, and this protocol we call Target2DeNovoDrugPropMax. This is followed by the fast automated AutoDock-Vina based in silico modeling and profiling of theinteraction of optimized drug leads and the drug target. This is followed by an automated execution of the Molecular Dynamics protocol that is also carried out for the complex identified with the best protein-ligand interaction from the AutoDock- Vina based virtual screening. The results are stored in the working folder of the user. The code is maintained, supported, and provide for use in thefollowing GitHub repositoryhttps://github.com/bengeof/Target2DeNovoDrugPropMaxAnticipating the rise in the use of quantum computing and quantum machine learning in drug discovery we use the Penny-lane interface to quantum hardware to turn classical Keras layers used in our machine/deep learning models into a quantum layer and introduce quantum layers into our classical models to produce a quantum-classical machine/deep learning hybrid model of our tool and the code corresponding to the same is provided belowhttps://github.com/bengeof/QPoweredTarget2DeNovoDrugPropMax

Network data is composed of nodes and edges. Successful application of machine learning/deeplearning algorithms on network data to make node classification and link prediction have been shownin the area of social networks through which highly customized suggestions are offered to socialnetwork users. Similarly one can attempt the use of machine learning/deep learning algorithms onbiological network data to generate predictions of scientific usefulness. In the presented work,compound-drug target interaction network data set from bindingDB has been used to train deeplearning neural network and a multi class classification has been implemented to classify PubChemcompound queried by the user into class labels of PBD IDs. This way target interaction prediction forPubChem compounds is carried out using deep learning. The user is required to input the PubChemCompound ID (CID) of the compound the user wishes to gain information about its predictedbiological activity and the tool outputs the RCSB PDB IDs of the predicted drug target interaction forthe input CID. Further the tool also optimizes the compound of interest of the user toward druglikeness properties through a deep learning based structure optimization with a deep learning baseddrug likeness optimization protocol. The tool also incorporates a feature to perform automated InSilico modelling for the compounds and the predicted drug targets to uncover their protein-ligandinteraction profiles. The program is hosted, supported and maintained at the following GitHubrepositoryhttps://github.com/bengeof/Compound2DeNovoDrugPropMaxAnticipating the rise in the use of quantum computing and quantum machine learning in drug discovery we usethe Penny-lane interface to quantum hardware to turn classical Keras layers used in our machine/deeplearning models into a quantum layer and introduce quantum layers into classical models to produce aquantum-classical machine/deep learning hybrid model of our tool and the code corresponding to thesame is provided belowhttps://github.com/bengeof/QPoweredCompound2DeNovoDrugPropMax

The article aims to analyze and evaluate the viability of the salvific hope promised by science and technology which has captured the modern mind. Traditionally the hope of salvation from the wretched human condition and passing into eudaimonic bliss hasbeen offered by religious figures and ideologies. However,the dominant idea of our times is that science and technology in its full growth can fix all human problems: poverty, disease, hunger, etc. It is pointed out that even though science and technology maybe able to solve the problems of disease and hunger, it is not adequate enough to answer our moral dilemmas or illumine our understanding of the meaning and purpose of life. Article concludes with the thoughts of S. Radhakrishnan (an Indian philosopher and statesman) that science and technology only provide the means to the ends proposed b yphilosophy or religion.

The on-going data-science and Artificial Intelligence (AI) revolution offer researchers a fresh set of tools to approach structure-based drug design problems in the computer-aided drug design space. A novel programmatic tool that incorporates in silico and deep learning based approaches for de novo drug design for any target of interest has been reported. Once the user specifies the target of interest in the form of a representative amino acid sequence or corresponding nucleotide sequence, the programmatic workflow of the tool generates compounds from the PubChem ligand library and novel SMILES of compounds not present in any ligand library but are likely to be active against the target. Following this, the tool performs a computationally efficient In-Silico modeling of the target and the newly generated compounds and stores the results of the protein-ligand interaction in the working folder of the user. Further, for the protein-ligand complex associated with the best protein-ligand interaction, the tool performs an automated Molecular Dynamics (MD) protocol and generates plots such as RMSD (Root Mean Square Deviation) which reveal the stability of the complex. A demonstrated use of the tool has been shown with the target signatures of Tumor Necrosis Factor-Alpha, an important therapeutic target in the case of anti-inflammatory treatment. The future scope of the tool involves, running the tool on a High-Performance Cluster for all known target signatures to generate data that will be useful to drive AI and Big data driven drug discovery. The code is hosted, maintained, and supported at the GitHub repository given in the link below https://github.com/bengeof/Target2DeNovoDrug Communicated by Ramaswamy H. Sarma