Prashanth Athri

Transfer RNAs (tRNA) are non-coding RNAs. Encouraged by biological applications discovered for peptides derived from other non-coding genomic regions, we explore the possibility of deriving epitope-based vaccines from tRNA encoded peptides (tREP) in this study. Epitope-based vaccines have been identified as an effective strategy to mitigate safety and specificity concerns observed in vaccine development. In this study, we explore the potential of tREP as a source for epitope-based vaccines for virus pathogens. We present a computational workflow that uses verified data sources and community-validated predictive tools to produce a ranked list of plausible epitope-based vaccines starting from tRNA sequences. The top epitope, bound to the predicted HLA molecule, for the virus pathogen is computationally validated through 200 ns molecular dynamics (MD) simulations followed by binding free energy calculations. The simulation results indicate that two tRNA encoded epitope-based vaccines, RRHIDIVV and IMVRFSAE for Mamastrovirus 3 and Norovirus GII, respectively, are likely candidates. Peptides originating from tRNAs provide unexplored opportunities for vaccine design. Encouraged by our previous experimental study, which established the inhibitory properties of tREPs against infectious parasites, we have proposed a computationally validated set of peptides derived from tREPs as vaccines for viral pathogens.

Background: RNA sequencing (RNA-Seq) is a technique that utilises the capabilities of next-generation sequencing to study a cellular transcriptome i.e., to determine the amount of RNA at a given time for a given biological sample. The advancement of RNA-Seq technology has resulted in a large volume of gene expression data for analysis. Results: Our computational model (built on top of TabNet) is first pretrained on an unlabelled dataset of multiple types of adenomas and adenocarcinomas and later fine-tuned on the labelled dataset, showing promising results in the context of the estimation of the vital status of colorectal cancer patients. We achieve a final cross-validated (ROC-AUC) Score of 0.88 by using multiple modalities of data. Conclusion: The results of this study demonstrate that self-supervised learning methods pretrained on a vast corpus of unlabelled data outperform traditional supervised learning methods such as XGBoost, Neural Networks, and Decision Trees that have been prevalent in the tabular domain. The results of this study are further boosted by the inclusion of multiple modalities of data pertaining to the patients in question. We find that genes such as RBM3, GSPT1, MAD2L1, and others important to the computation model’s prediction task obtained through model interpretability corroborate with pathological evidence in current literature.

Molecular dynamics (MD) simulations probe the conformational repertoire of macromolecular systems using Newtonian dynamic equations. The time scales of MD simulations allow the exploration of biologically relevant phenomena and can elucidate spatial and temporal properties of the building blocks of life, such as deoxyribonucleic acid (DNA) and protein, across microsecond (μs) time scales using femtosecond (fs) time steps. A principal bottleneck toward extending MD calculations to larger time scales is the long-range electrostatic force measuring component of the naive nonbonded force computation algorithm, which scales with a complexity of (N, number of atoms). In this review, we present various methods to determine electrostatic interactions in often-used open-source MD packages as well as the implementation details that facilitate acceleration of the electrostatic interaction calculation.

The ability to estimate the probability of a drug to receive approval in clinical trials provides natural advantages to optimizing pharmaceutical research workflows. Success rates of clinical trials have deep implications to costs, duration of development, and under pressure due to stringent regulatory approval processes. We propose a machine learning approach that can predict the outcome of the trial with reliable accuracies, using biological activities, physicochemical properties of the compounds, target-related features, and NLP-based compound representation. Biological activities have never been used as a predictive feature. We have extracted the drug-disease pair from clinical trials and mapped target(s) to that pair using multiple data sources. Empirical results demonstrate that ensemble learning outperforms independently trained, small-data ML models. We report results and inferences derived from a Random forest classifier with an average accuracy of 93%, and an F1 score of 0.96 for the 'Pass' class. 'Pass' refers to one of the two classes (Pass/ Fail) of all clinical trials and the model performed well in predicting the 'Pass' category. An analysis of the features demonstrates that bioactivity plays an important role in predicting the outcome of a clinical trial. A significant effort has gone into the production of the dataset that, for the first time, integrates clinical trial information with protein targets. All code to map these entities is available through this study, and all data are from publicly available sources. While our model identifies low-lying inferences when biological activities are included, the code to integrate biological activity and target information provide researchers with access to deep curated and proprietary clinical trial databases the ability to get deeper insights, better statistical significance, and capabilities to better predict trial failures.

The ability to estimate the probability of a drug to receive approval in clinical trials provides natural advantages to optimizing pharmaceutical research workflows. Success rates of clinical trials have deep implications for costs, duration of development, and under pressure due to stringent regulatory approval processes. We propose a machine learning approach that can predict the outcome of the trial with reliable accuracies, using biological activities, physicochemical properties of the compounds, target‐related features, and NLP‐based compound representation. In the above list, biological activities have never been used as an independent variable towards the prediction of clinical trial outcomes. We have extracted the drug–disease pair from clinical trials and mapped target(s) to that pair using multiple data sources. Empirical results demonstrate that ensemble learning outperforms independently trained, small‐data ML models. We report results and inferences derived from a Random forest classifier with an average accuracy of 93%, and an F1 score of 0.96 for the “Pass” class. “Pass” refers to one of the two classes (Pass/Fail) of all clinical trials, and the model performed well in predicting the “Pass” category. Through the analysis of feature contributions to predictive capability, we have demonstrated that bioactivity plays a statistically significant role in predicting clinical trial outcome. A significant effort has gone into the production of the dataset that, for the first time, integrates clinical trial information with protein targets. Cleaned, organized, integrated data and code to map these entities, created as a part of this work, are available open‐source. This reproducibility and the freely available code ensure that researchers with access to deep curated and proprietary clinical trial databases (we only use open‐source data in this study) can further expand the scope of the results.

Adverse drug reactions (ADRs) are pharmacological events triggered by drug interactions with various sources of origin including drug–drug interactions. While there are many computational studies that explore models to predict ADRs originating from single drugs, only a few of them explore models that predict ADRs from drug combinations. Further, as far as we know, none of them have developed models using transcriptomic data, specifically the LINCS L1000 drug‐induced gene expression data to predict ADRs for drug combinations. In this study, we use the TWOSIDES database as a source of ADRs originating from two‐drug combinations. 34,549 common drug pairs between these two databases were used to train an artificial neural network (ANN), to predict 243 ADRs that were induced by at least 10% of the drug pairs. Our model predicts the occurrence of these ADRs with an average accuracy of 82% across a multifold cross‐validation.

Adverse drug reactions (ADRs) are pharmacological events triggered by drug interactions with various sources of origin including drug-drug interactions. While there are many computational studies that explore models to predict ADRs originating from single drugs, only a few of them explore models that predict ADRs from drug combinations. Further, as far as we know, none of them have developed models using transcriptomic data, specifically the LINCS L1000 drug induced gene expression data to predict ADRs for drug combinations. In this study we use the TWOSIDES database as a source of ADRs originating from two-drug combinations. 34,549 common drug pairs between these two databases were used to train an artificial neural network (ANN), to predict 243 ADRs that were induced by at least 10% of the drug pairs. Our model predicts the occurrence of these ADRs with an average accuracy of 82% across a multi fold cross validation. Source Code and input dataset used in this study can be found at: https://bitbucket.org/ishita98/prediction-of-adr/src/master/

We compare various Machine Learning and Deep Learning algorithm variants and string embeddings to predict the pIC50 values against the JAK2 protein target. The paper works upon improving the existing predictor component in the ReLeaSE framework by comparing various models such as Random Forest Regressor and Convolutional Neural Network for prediction. The paper also works on using different SMILE string embeddings such as PPMI matrices and OpenBabel fingerprints for molecules to improve upon the current baseline accuracy. Using a data set consisting of an approximate of 2000 data points, the models were trained on 80:20 split. The comparison of various models provides us with insights on the robustness in using SMILES as a representation for molecules. The results show that the use of Random Forest Regressor along with OpenBabel fingerprinting to encode SMILES performed the best with an R ² score of 0.71, which improves the baseline predictor in ReLeaSE by a ΔR ² of 23%. Our models when trained on an augmented version of the JAK2 dataset gave an R ² score of 0.97

Adaptive sampling molecular dynamics based on Markov State Models use short parallel MD simulations to accelerate simulations, and are proven to identify hidden conformers. The accuracy of the predictions provided by it depends on the features extracted from the simulated data that is used to construct it. The identification of the most important features in the trajectories of the simulated system has a considerable effect on the results. Methods: In this study, we use a combination of Laplacian scoring and genetic algorithms to obtain an optimized feature subset for the construction of the MSM. The approach is validated on simulations of three protein folding complexes, and two protein ligand binding complexes. Results: Our experiments show that this approach produces better results when the number of samples is significantly lesser than the number of features extracted. We also observed that this method mitigates over fitting that occurs due to high dimensionality of large biosystems with shorter simulation times.

Computational approaches to analyze various drug/ compound centered analysis often present a need to map attributes from multiple drug databases. In this study, we provide a Neo4j repository that integrates two of the most prominent open source drug databases, DrugBank and ChEMBL, with a goal of establishing an integrated data visualization and analysis tool for drug discovery studies. The drugs present in DrugBank are mapped to their counterparts in ChEMBL. The integration of these resources and the harmonization using knowledge graph serialization using Neo4j lead to identification of relationships between drugs and other related features that are otherwise spread across two different resources. A common data format, a prerequisite to populate the Neo4j database, enables users to identify new relationships central to drug discovery research, like Drug Target Interactions (DTI). The resource is freely available at: https://github.com/ambf0632/CompoundDB4j