Research in Saha lab involves development and application of novel computational chemistry and biophysics approaches to gain fundamental understanding of remarkable biological processes and their relation to complex diseases. This knowledge will further be used to design small molecules either to inhibit or to enhance particular biological function facilitating drug discovery for cancer and neurodegenerative diseases (e.g., Alzheimer’s, Parkinson’s, Dementia). To achieve this goal, state-of-the-art computer aided drug discovery techniques will be implemented in collaboration with medicinal chemistry, biochemistry and experimental biophysical research groups (both in academia and in pharmaceutical industry). Researchers in Saha lab will develop skills and expertise in advanced molecular modeling techniques (e.g., Machine Learning, Coarse grained Molecular Dynamics, Quantum Mechanics, Molecular Mechanics, Cheminformatics) as well as simulated exposure to computational drug discovery research as in Big Pharma.

Current Projects

Project 1: Drug Discovery for Protein Protein Interfaces

Protein-protein interactions (PPI) are considered undruggable targets for drug discovery. However, more than half a million PPI have been suggested to be the part of the human interactome representing an enormous potential for drug discovery. Drug discovery for PPIs severely suffers from the lack of adequate tools/knowledge to understand the highly dynamic nature of PPI interface.

Our objective in this project is to develop a new method that combines machine learning and computational biophysics to predict peptides and small molecules to inhibit new and challenging PPIs. Our current focus is to inhibit the complex between interleukin receptors IL-4Ra and IL-13Ra1 and interleukin IL-4. This interaction is upregulated for chronic skin disease atopic dermatitis (AD)/eczema.
Project 2: Covalent Drug Discovery through Computational Enzymology

Covalent inhibitors are seeing renewed interest in recent times. Despite having toxicity issues (originating from highly non-specific nature in binding), covalent drugs offer several benefits such as targeting “undruggable” proteins, high ligand efficiency and potential to fight resistant mutations. Unlike non-covalent drugs, discovery of covalent drugs suffers from lack of knowledge in underlying catalytic reaction mechanisms.

Our goal in this project is to develop robust computational enzymology framework to design effective reversible covalent inhibitors and enhancers (i.e. PROTACs) with good pharmacokinetic properties and high selectivity. Our current focus is to develop efficient inhibitors for bone marrow cancer (i.e., Multiple Myeloma). We will simultaneously employ these models to gain molecular level understanding of the cascade of enzymatic reactions in Ubiquitin Proteasome Pathway that are responsible for different types of cancer and neurodegenerative diseases.Project 3: Understanding Mechanochemical Processes in Motor Proteins

Protein-based Molecular Motors (aka biological motors or motor proteins) are nature’s solution to efficient conversion of chemical energy to mechanical movement in maintaining the cell’s cargo activity. Proteasome is a fascinating biological motor that uniquely identifies damaged or obsolete regulatory proteins and degrades them into smaller peptides and hence defects in proteasome are broadly implicated in severe diseases including cancer and neurodegenerative disorders. Mechanism of how protein is processed and subsequently degraded inside proteasome is not known in molecular detail.

With the emerging Cryo-EM structures of the proteasome, our goal here is to (1) Develop novel high resolution multiscale models (HR-MM) with precise hierarchical layering to study multi-subunit protein complexes. (2) Employ these models to gain fundamental insights into the coupling of different physical and chemical mechanisms that propels the sophisticated mechanochemical processes inside proteasome. (3) Determine the role of structural landscape and chemical intricacies of each domain in defining the unique functional pathway of this motor.

Project 4: Combining Medicinal Chemistry and Machine Learning in Drug Design

Traditional drug discovery relies on screening of billions of small molecules both virtually and experimentally requiring vast amount of resources. Despite the tremendous potential/application of machine learning models, this “Needle in a haystack” problem still remains as a bottleneck.

Our goal in this project is to develop a new drug design algorithm by combining machine learning models with strong knowledge of chemistry in bioactive space and building on existing principles of fragment based drug discovery. Our current focus is to revisit bruton tyrosine kinase inhibitor landscape to explore this proof of concept and further extend this tool to design efficient inhibitors for challenging targets such as RNA-binding proteins (RBPs).