Shashank Pant was a graduate student in the Center for Biophysics and Quantitative Biology at the University of Illinois Urbana-Champaign. 

He was awarded the Beckman Fellowship by the Beckman Foundation University of Illinois in 2019. 

INDIAai interviewed Shashank to get his perspective on AI.

A chemical science graduate becomes a computational biophysics/chemistry scientist. How did it all begin? Tell us about your journey.

The field of computational chemistry/biophysics is a highly multidisciplinary area. It exploits chemical sciences, biology, mathematics, and statistical physics methods, incorporated into efficient computational codes to unravel the function of proteins and molecules at an atomic resolution. 

During my undergraduate studies, I became interested in chemical sciences, specifically polymer physics. I started my research career as an undergraduate researcher at the Polymer Research Center at IISERK. In the lab of Dr P. De, I designed and developed pH and thermoresponsive polymers for drug delivery. These polymers assemble into higher-order structures sensitive to thermal and acidic stress. Over two years as an undergraduate researcher, I uncovered that the origin of higher-order assemblies is attributed to the balance between the hydrophobic and hydrophilic chemical groups. This phenomenon is known as the "hydrophobic effect". I took multiple statistical physics and chemical sciences courses to understand this phenomenon better. Finally, I was fortunate enough to receive a National Initiative for Undergraduate Sciences (NIUS) fellowship to research this topic. This fellowship provided me with a platform to work on the origin of hydrophobicity at the nanoscale under the able guidance of Dr Niharendu Choudhury at Bhabha Atomic Research Center (BARC). Using simplistic computational models, I understood how polar molecules, like water, arranged themself into ordered assemblies near hydrophobic surfaces. After completing this project, I got opportunities to pursue summer research in similar research areas at the University of Texas and the University of Manchester. The interdisciplinary nature of these projects attracted me towards the area of computational & theoretical chemistry and biophysics. During these summer research internships, I applied my mathematics, computer sciences, physics, chemistry, and biology knowledge to solve complex problems. Additionally, my graduate days at the University of Illinois Urbana-Champaign gave me more opportunities to acquire biology and artificial intelligence skills and establish myself as a computational biophysicist. 

What motivated you to use AI for complex biological processes?

The field of AI has roots in almost every branch of research, including physics, mathematics, computing, psychology, and biology. In terms of applications and themes, AI encompasses topics ranging from perception, cognition, and automation to drug discovery and other biological processes. In this current era, we have to be open-minded and creative when it comes to employing AI. Yet, surprisingly, some people still have a very rigid view of AI. 

My undergraduate studies at the Indian Institute of Science Education and Research Kolkata (IISERK) helped lay my solid fundamental foundation in science. One exciting thing about IISERs is that it allows you to choose any subject/topic as a minor, allowing me to take courses like statistical physics, linear algebra, graph theory, computing, and biology, along with majors in chemical sciences. After moving to the University of Illinois, Urbana-Champaign, I became aware of rapid advancements in AI and how it can improve conventional techniques in computational biology. Thanks to my PhD advisor, Dr Emad Tajkhorshid, I got a chance to do summer research in the lab of Dr Pratyush Tiwary at the University of Maryland. During this time, I learned and combined statistical physics, AI, and biology to accelerate molecular simulations and increase the trustworthiness of these methods in the data sparse regime. It was an extremely satisfying research experience because I could appreciate the power of AI when used intelligently. 

Tell us about your PhD work and the contributions you made to it.

My research focuses on developing and employing synergies between statistical mechanics, computer simulations and artificial intelligence (AI) to study the kinetics and thermodynamics of complex biological processes. I am specifically interested in performing all-atom simulations that can reach milliseconds and longer timescales to bridge the gap between theory and experiments. As these timescales are far beyond the reach of the fastest supercomputers, my work entails using a variety of enhanced sampling algorithms to understand the dynamics of biomedically relevant biomolecules.  

In my doctoral research, I focused on combining statistical mechanics, computer simulations and artificial intelligence (AI) to unravel the conformational landscape of membrane proteins, hoping to develop novel drug design strategies. The overall estimate of the human genome suggests that membrane proteins make up 25% of the human proteome. They intimately interact with biological membranes, thus allowing for controlled and selective processing of information reaching the cells. Pharmacologically, these proteins play critical roles in drug absorption, distribution, and elimination and are even drug targets themselves. Thus, due to its fundamental role in diverse biological and physiological processes, membrane proteins are among the critical drug targets in various human disorders, further stimulating widespread interest in mechanistic studies of these proteins at an atomic level. Due to the vital importance of membrane proteins in biology and medicine, and the recent extensive investment in their structural biology, the past few years have witnessed an astonishing number of high-resolution structures of these proteins.

Nevertheless, we still have a minimal view of the nature of the structural transitions between the central functional states and the coupling of these changes to other molecular events. With the overall increase in the burden of neurological disorders and cancer worldwide, my research helped lay the foundation for developing novel therapeutic strategies to target various membrane proteins involved in these disorders. During my doctoral research, I developed computational tools, algorithms, and methods that have played a critical role in providing a platform for computational researchers and researchers in other fields, such as structural biology and electrophysiology.

The second part of my doctoral research dealt with the importance of biological membranes in dictating the structure and function of membrane proteins. My research highlighted how signalling lipids play a significant role in the mediating opening of ion channels, an important drug target for neurological disorders, and assists physicians in identifying what kinds of anti-epileptic drugs can be used based on the mutations. In addition, it will open the area of personalized diagnoses and medicine for neonatal epilepsy.

What initial challenges did you encounter as a researcher in India and the United States? Could you describe how you overcame them?

I came to the United States in 2016 after completing my integrated MS from IISERK. Due to my past research experience as a summer research student in the UK and US, I hardly faced any difficulties regarding research culture. However, students usually face challenges adjusting to the culture, accent, time management, academics, and overall social scene. My advice is, "Don't be Shy, bold, and confident. If you can think it, you can do it!" 

How did you respond to disappointments like experiment failures and paper rejection as a researcher? How do you keep your composure?

There is no such thing as success without failures and mistakes. The most satisfying thing in research is the process of constant feedback that you get from your peers, which helps you improve as a scientist. This feedback can be in the form of questions asked during conference presentations and comments/rejections during peer reviews. I often incorporate these comments (primarily constructive) into my research which strengthens my work. As far as unsuccessful experiments are concerned, I constantly analyze experiments to identify the reasons behind the failure, which helps in highlighting the weakness and forces you to think out-of-the-box to improve it and make it widely applicable. Research is a journey which should be enjoyed and shouldn't be driven just based on milestones or success stories. 

The Beckman Foundation University of Illinois awarded you a Beckman Fellowship in 2019. Tell us about the Beckman Fellowship. What steps should researchers take to apply for such fellowships?

The Beckman Institute is a physical space and community of researchers to foster cross-platform collaborations and nurture new ideas and discoveries. The sole purpose of this institute is to reduce the threshold/barriers between conventional research and cutting-edge technologies to yield research breakthroughs. To encourage collaborations, Beckman Institute offers a Graduate Fellowship Program, which provides a platform for University of Illinois graduate students to pursue interdisciplinary research at the institute. This program is supported by funding from the Arnold and Mabel Beckman Foundation. 

Each applicant must submit a proposal for an interdisciplinary research project involving University of Illinois faculty members from different research groups. All the applications are reviewed solely regarding the proposed work's quality, including the development of novel technologies. Selected applicants for this program are hired as graduate research assistants. I received this fellowship in 2019 for my research project on unravelling the structure and function of epilepsy-causing ion channels. My advice to current and incoming students, specifically international students, is to keep an eye out for these limited fellowships, plan well in advance, and be original & aggressive in writing proposals.

What advice would you provide to someone considering a career in artificial intelligence research? What should they concentrate on to advance?

Someone considering a career in AI should know that AI is powerful and easy to use. But, at the same time, it is limited and easy to misuse and will not replace intuition and physical laws.

As far as advice is concerned, I recommend the following.

1. Firstly, build a strong foundation in AI/ML. I believe linear algebra and probability are two main topics necessary to appreciate and understand current state-of-the-art AI/ML algorithms. 

2. Understand the strengths and limitations of AI/ML techniques. It might allow researchers to make well-informed decisions in choosing a specific algorithm to solve the problem.

3. Think deeply!

What academic works and publications have had the most significant influence on your life?

The following articles and books on statistical physics and AI/ML have influenced my research. 

Statistical physics and thermodynamics:

1. "Statistical Mechanics: Theory and Molecular Simulation" by Mark E. Tuckerman

2. "Introduction to Modern Statistical Mechanics" by David Chandler  

3. "An Introduction to Statistical Thermodynamics" by Terrell L. Hill

4. "Nonequilibrium Statistical Mechanics" by Robert Zwanzig

5. "Statistical Mechanics" by Donald Allan McQuarrie

6. Nonequilibrium equality for free energy differences, C Jarzynski, Physical Review Letters 78 (14), 2690

7. Entropy production fluctuation theorem and the nonequilibrium work relation for free energy differences, GE Crooks, Physical Review E 60 (3), 2721.

8. The protein-folding problem, 50 years on, KA Dill, JL MacCallum, Science 338 (6110), 1042-1046

9. Role of repulsive forces in determining the equilibrium structure of simple liquids, JD Weeks, D Chandler, HC Andersen, The Journal of chemical physics 54 (12), 5237-5247

10. Interfaces and the driving force of hydrophobic assembly, D Chandler, Nature 437 (7059), 640-647

Below is an incomplete list of books that have helped me a lot in understanding AI

1. "Deep Learning" by Aaron Courville, Ian Goodfellow, and Yoshua Bengio 

2. "Elements of Information Theory" by Joy Thomas and Thomas Cover

3. Roughgarden T., Twenty lectures on algorithmic game theory. Cambridge University Press; 2016