Palash Bera

This study evaluates the performance of a generative diffusion model in sampling protein conformations across different sizes and structures, showing it can reproduce key features but has limitations in capturing low-populated conformers.

This paper shows that a transformer model can predict future states of physicochemical systems faster than traditional simulations by learning from time-discretized molecular dynamics data.

This study shows that protein synthesis, particularly ribosome dynamics, enhances fluidity in bacterial cytoplasm by counteracting its glassy nature through metabolic activity.

This paper uses machine learning to analyze Hi-C data from E. Coli, revealing structural patterns and predicting DNA dynamics, showing that a subset of contact pairs suffices for accurate predictions.