Biochemical Graph Reconstruction and Representation using AI/ML Techniques

Abstract

Modeling of biochemical systems as a network is the central theme in network biology. The network representation of biochemical systems leads to biochemical graphs. These graphs can be used either as models for further analysis or a data set for building data-driven models for different purpose. In this talk, we discuss problems related to reconstruction and representation of biochemical graphs. First, we present the problem of reconstructing biological networks from the literature data using deep learning models. Then, we discuss an approach to represent molecular graphs using the concept of functional groups. The application of this new representation in property predictions is presented and is compared with DMPNN, a state of art deep learning model. We conclude the talk with other interesting problems in the interface of biochemical graphs and AI.

Biography of the speaker

Dr. Nirav Bhatt is an an Assistant Professor at the Department of Biotechnology, Bhupat and Jyoti Mehta School of Biosciences. He is faculty-in-Charge pCoE for Network Systems Learning, Control and Evolution , core faculty member at Robert Bosch Centre for Data Science and Artificial Intelligence (RBCDSAI) and Integrative Biology and Systems MedicineE (IBSE), IIT Madras. His research interests include first-principle and machine learning approaches for modeling and control of biological networks from multi-sensor and multi-scale data, safe learning and control of man-made networks, and bioprocess data analysis for process monitoring. He earned MTech from IIT Madras and PhD from EPFL, Switzerland.