Focus_Areas

Focus Areas

Our Major Research Areas

AI/ML Applications

Artificial Intelligence for Applications in Chemistry, Drug Discovery, Biology, and Healthcare

Dataset Development

Developing Large Scale Datasets for Machine Learning Applications in Drug Discovery and Healthcare

Biomolecular Systems

Molecular Dynamics Simulations, State of the Art Enhanced Sampling Techniques and QM/MM for Studying Biomolecular Processes

Computational Materials & Reaction Pathways

Quantum Chemical Calculations for Investigating Electronic Properties, Reaction Profiles of Chemical Processes

AI/ML Applications

Applications of AI/ML is one of our focus areas with applications in chemistry, biology and healthcare. We use state of the art methods such as recurrent neural network (RNN), convolutional neural network (CNN), reinforcement learning (RL), generative pre-trained transformers (GPT), generative adversarial network (GAN), variational autoencoder (VAE), diffusion models for molecule/material design. Some of the problems we have addressed are generation of new molecules/materials using GenAI algorithms, retrosynthesis, spectra to structure elucidation, RNA structure prediction, drug-drug interaction, protein-ligand interactions, machine learning force field, etc. We are also interested in developing deep learning architectures that incorporates chemical interpretability.

We also work on problems in healthcare such as development of prognostic ML models for infectious diseases, image/signal processing for diagnosis of pulmonary/cardiovascular diseases.Given the variability of healthcare data, class imbalances, dataset shift, etc., we also focus on self-supervised learning approaches and various explainable methods. In collaboration with Dr. Maitreya Maiti, we develop biomedical devices/wearables.

Dataset Development

Modern AI/ML are powerful and have the potential to make the drug discovery process efficient.However, the generalizability and reliability of these methods depend on the data availability. We are in the process of developing the largest ever synthetic dataset on protein-ligand complexes that will drive method development tackling different stages of the drug design pipeline. We have already released PLAS-5k, PLAS-20k, and APOBIND datasets. PLAS-20k dataset is a result of 100,000 MD simulations on a total of 20,000 protein-ligand complexes wherein the trajectory data and the MM-PBSA free energy components have been made available.
We work with intel, AWS and in silico medicine in these efforts.

Biomolecular Systems

We use molecular dynamics simulations (mostly using the CHARMM force fields) to investigate fundamental processes in biological systems. We have significantly contributed in the areas of RNA folding dynamics, protein folding equilibrium shifts due to chemical/mechanical/thermal perturbations, ion channels, transporters, hybrid nucleic acid systems for applications in anti-sense therapy, transmembrane protein modeling, protein-DNA interactions, etc.
Our group is also involved in computational drug discovery through protein-ligand binding affinity prediction/estimation and mechanism and end point binding free energy calculations, implementing molecular mechanics Poisson Boltzmann Surface Area (MMPBSA) method.
Additionally, we use MD simulations to study chemical systems such as interfacial dynamics polymer protected nanoparticle systems.

Computational Materials and Reaction Pathways

Semi-empirical calculations are a category of quantum chemical calculations used to estimate the electronic structure and properties of molecules. They strike a balance between accuracy and computational efficiency by employing a set of empirical parameters fitted to experimental data. Unlike ab initio methods that rigorously solve the Schrödinger equation, semi-empirical methods simplify calculations by introducing empirical adjustments for certain terms, making them computationally less demanding. These methods are often used for larger molecular systems where the exhaustive accuracy of ab initio methods may be impractical.

Our group has studied oxidation of greenhouse gas CO on metal cluster, bench marked various quantum mechanical methods to model small chemical reactions and has studied the role of basis set to capture molecular properties in cyclic systems. Electronic structure analysis aid in understanding experimental result, and our group has studied photocatalytic reactions, functionalization of cyclic systems, in collaboration with experimental groups. Our group also focuses on the dynamics of tetrahedral centres to mark the first study of stereochemical inversion without bond breaking.