Doctoral Student

PhD Position in Physics-Informed Graph Convolutional Learning for Optimization and Dynamics of Biochemical Reaction Networks

Institution:

The Amrita School of Artificial Intelligence, a constituent school of Amrita Vishwa Vidyapeetham, is a multidisciplinary, research-intensive institution located at Ettimadai, Coimbatore, Tamil Nadu, India. The school is committed to advancing the theory and practice of Artificial Intelligence through high-quality education and cutting-edge interdisciplinary research spanning engineering, science, healthcare, and computational biology.

Selected PhD scholars will receive a monthly fellowship of ₹30,000 during the initial stage of the program, along with access to modern computational facilities, high-performance computing resources, and collaborative research opportunities with leading academic institutions, including the IITs. The doctoral program has a duration of three years. Amrita Vishwa Vidyapeetham strongly emphasizes ethical, socially responsible, and interdisciplinary AI research that addresses real-world scientific and societal challenges.

Position Description:

Applications are invited for a full-time, on-campus PhD position at the Amrita School of Artificial Intelligence, Coimbatore, under the research project:

Physics-Informed Graph Convolutional Learning for Optimization and Dynamics of Biochemical Reaction Networks

The selected doctoral candidate will conduct original research at the intersection of Artificial Intelligence, Graph Neural Networks, Systems Biology, and Nonlinear Dynamical Systems. The research activities will include:

• Conducting comprehensive literature reviews and identifying emerging research directions.
• Developing novel physics-informed graph learning algorithms for biochemical reaction networks.
• Designing computational models and implementing algorithms using modern AI frameworks.
• Performing numerical simulations, benchmarking, and validation using biological datasets.
• Publishing research findings in leading international journals and conferences.
• Participating in seminars, workshops, and collaborative research activities with national and international research groups.
• Supporting undergraduate and postgraduate research projects and, where appropriate, assisting with teaching activities.

The position offers an excellent opportunity to work in a highly collaborative research environment with faculty members and researchers from multiple disciplines.

Project Description:

Biochemical reaction networks govern a wide range of fundamental biological processes, including metabolism, cellular signaling, gene regulation, enzyme kinetics, and protein interaction networks. These systems exhibit highly nonlinear, multiscale, and interconnected dynamics that are often difficult to analyze using conventional mathematical models and computational techniques. While traditional systems biology approaches and kinetic modeling have significantly advanced our understanding of biochemical processes, they frequently encounter limitations when addressing large-scale biological networks, noisy experimental observations, incomplete reaction mechanisms, and parameter uncertainty.

Recent advances in Artificial Intelligence, particularly Graph Neural Networks (GNNs) and Graph Convolutional Networks (GCNs), have created new opportunities for modeling complex biological systems. Since biochemical reaction systems naturally possess graph structures—with molecular species represented as nodes and biochemical interactions represented as edges—graph-based deep learning provides a powerful framework for extracting structural and dynamical information from biological networks. However, conventional graph learning methods are primarily data-driven and often lack physical interpretability. Consequently, they may violate fundamental biochemical principles, including mass conservation, reaction kinetics, thermodynamic constraints, and nonlinear dynamical behavior. These limitations motivate the development of physics-informed graph learning frameworks that explicitly incorporate mechanistic knowledge into deep learning architectures.

This research project aims to develop a novel Physics-Informed Graph Convolutional Learning (PI-GCL) framework for analyzing, predicting, and optimizing biochemical reaction networks. The proposed framework will integrate the representation-learning capability of Graph Convolutional Networks with mechanistic models derived from biochemical reaction kinetics and nonlinear dynamical systems. Physics-informed constraints—including reaction-rate equations, conservation laws, thermodynamic principles, and governing differential equations—will be embedded directly into the learning process. By combining data-driven learning with domain-specific physical knowledge, the proposed framework is expected to significantly improve the accuracy, robustness, stability, and interpretability of biochemical network models.

The developed methodology will enable:

• Learning biologically meaningful representations of biochemical networks.
• Identification of critical reaction pathways and regulatory mechanisms.
• Optimization of metabolic and signaling pathways.
• Prediction of emergent system behavior under noisy and uncertain conditions.
• Discovery of key molecular interactions that govern complex biological processes.

The proposed framework will be evaluated using benchmark datasets from systems biology, including metabolic networks, signaling pathways, gene regulatory networks, and protein interaction networks.

The project is expected to make significant contributions at the intersection of Artificial Intelligence, Graph Neural Networks, Nonlinear Dynamical Systems, and Computational Systems Biology. Its outcomes have potential applications in disease modeling, drug target identification, synthetic biology, metabolic engineering, precision medicine, and AI-driven biological discovery.

Eligibility:

Applicants with any of the following educational backgrounds are encouraged to apply:

• M.Sc. in Physics
• M.Sc. in Physical Chemistry
• M.Sc. in Applied Mathematics
• M.Sc. in Computer Science
• MCA
• M.Tech./B.E./B.Tech. in Mechanical Engineering or related disciplines

Desired Qualifications:

Applicants should possess:

• Strong programming skills in Python, MATLAB, or related scientific programming languages.
• Working knowledge of machine learning frameworks such as PyTorch, TensorFlow, or scikit-learn.
• A solid foundation in linear algebra, optimization, probability, statistics, differential equations, and numerical methods.
• The ability to conduct independent research, including literature review, problem formulation, algorithm development, experimental design, data analysis, and scientific interpretation.
• Excellent written and oral communication skills, with the ability to prepare high-quality research papers, technical reports, and presentations.
• A strong interest in Artificial Intelligence, Graph Neural Networks, Computational Biology, Scientific Machine Learning, or related interdisciplinary research areas.
• The ability to work full-time in a collaborative research environment while maintaining high standards of academic integrity and achieving long-term research milestones.

Prior research experience, publications, or industry projects related to Artificial Intelligence, Data Science, Machine Learning, Computational Biology, or Graph Learning will be considered an advantage but are not mandatory.