Balaraman Ravindran

PURE: Policy-guided Unbiased REpresentations for structure-constrained molecular generation

The paper introduces PURE, a method for structure-constrained molecular generation that avoids metric biases and achieves competitive performance through reinforcement learning and self-supervised learning.

A deep network-based model of hippocampal memory functions under normal and Alzheimer's disease conditions

The paper presents a deep network model of hippocampal memory functions, simulating normal and Alzheimer's disease conditions by altering network damage levels and observing recall behavior.

MultiCens: Multilayer network centrality measures to uncover molecular mediators of tissue-tissue communication

MultiCens identifies inter-tissue molecular mediators by analyzing multi-layer network centrality, revealing genes involved in hormone signaling and Alzheimer's disease brain regions.

Memory Consolidation with Orthogonal Gradients for avoiding Catastrophic Forgetting

The paper proposes a regularization-based model to prevent catastrophic forgetting in deep networks by constraining parameter updates to directions orthogonal to previous task gradients, showing improved performance on standard datasets.

NetGenes: A database of essential genes predicted using features from interaction networks

NetGenes predicts essential genes in 2700+ bacteria using network-based features from STRING databases and provides essentiality scores and annotations.

Sequence neighborhoods enable reliable prediction of pathogenic mutations in cancer genomes

This study develops NBDriver, a tool that uses sequence neighborhoods to predict pathogenic mutations in cancer genomes, outperforming existing methods in identifying driver mutations.