Rajesh Kumar, Senior Director of Structural Biology at Schrodinger, shares with Diana Turner how advanced computational models are supercharging modern drug discovery and the impact they could have in the future. 

DT: Can you tell us a bit more about Schrödinger and what you do?

RK: Schrödinger’s (SDGR) core objective is to empower science-driven companies around the world to meet their goals with greater efficiency and productivity, and to help academic and government research labs make important scientific discoveries aimed at improving human health and quality of life. We do this by building and deploying scientific software solutions and forming collaborations and partnerships with commercial and non-commercial research organisations. 

A few years ago, SDGR decided to initiate its own drug discovery programme and formed the Schrödinger Therapeutics Group (STG) that built a strong team of scientists with expertise in biology & genetics, chemistry, modeling, and structural biology (SB). As high-resolution atomic structures of liganded drug targets enable the predictive power of our computational platform, a lab focused on SB and biophysics was built in Framingham, a suburb of Boston. I lead a team focused on protein biochemistry, biophysics and SB of human drug targets. In the last couple of years, we have built a state-of-the art lab with the latest technology that will help us accelerate our experimental workflow and obtain very high-quality data that would help our internal projects portfolio 

DT: How do Schrödinger’s services aim to reduce the time and cost of research?

RK: In addition to our own internal drug discovery portfolio (with two assets in clinical trials), SDGR’s platform is a suite of physics-based programmes that help model both atomic level and macroscopic chemical interactions to study and predict a wide variety of physico-chemical behaviour of molecules. Our machine learning approaches create robust models utilising proprietary physics-based force fields, which then accelerates the discovery process by predicting about 10 best molecules from billions. Our experience over the years show that typically two to three of these molecules are top drug candidates. 

DT: How does Schrodinger use Sapio’s AI-powered Electronic Lab Notebook (ELN)?

RK: The volume of high-quality data generated by our SB and biophysics lab needs to be documented in a diligent and timely fashion ensuring compliance and ready availability of data to be shared with stakeholders within our organisation. Since implementing our SDGR specific ELN/LIMS, Sapio has been very quick and receptive to make any requested tweaks in the implementation, as well as helping us use the tool very efficiently. Recently, Sapio’s incorporation of AI tools as part of their platform (ELaiN) is helping us do regular queries on our data in a more seamless fashion. For example, I can list all experiments created by a user by different projects or plot them to obtain some visual metrics. This seems to be a powerful tool, and we will start using more of the AI based tool within our ELN/LIMS. SDGR has a very open culture and ready to test out and adapt new technologies and tools available from the literature or from our vendors/collaborators. With that same spirit, we will definitely adapt to Sapio’s AI tools to make our mundane tasks more automated. 

DT: What impact do you think AI has had on data capture and lab workflows across the industry as a whole?

RK: In the past year, a surge in AI tools that have become available to the common man and the scientific community is helping us with tasks that seemed very intense in the past. SDGR, one of the early adapters of machine learning tools, has our own set of AI assistants incorporated in our computational platform, to parse and make sense of large amounts of data. 

I am seeing a lot of other efficient tools coming on board in the past year that will help solve challenges in analysing large volumes of data and make sensible recommendations. I see large project meetings being effectively transcribed by Google’s Gemini tools with key discussion points and next steps. It is amazing to read the meeting reports generated by AI with pretty good accuracy. Recently I read that MIT is pioneering discovery of novel antibiotics using AI that will help patients with antibiotic resistance because of continuous use of the previous generation antibiotics. There is a lot of new tools appearing where AI can be trained to arrive at novel therapeutics across all modalities (ex: small molecule, biologics, etc.) Large public databases that carry very important patient derived, real time data (i.e. UK Biobank) can be easily parsed by AI to help steer towards more targeted therapies for patients. 

DT: How do you think combining AI and other advanced computational models will benefit drug discovery in the next five to 10 years?

RK: Science is built on hypothesis and predictions but in the end has to be proven beyond doubt by experimental facts. Using appropriate tools such as AI and advanced computational models can help drive speed, accuracy, quality and efficiency to come up with desired outcomes for the larger human community and patients suffering from dreadful diseases. Although not all AI models and platforms are on par with the growing needs, there is going to be an accelerated evolution of AI concepts and engines that we need to look out for in the next five to 10 years. The best of AI tools combined with the right data or models, can help us move towards a golden age of drug discovery. This gives us hope that we can soon treat dreadful diseases such as cancers, neurodegenerative and immunological diseases and deal with a rise in pathogenic diseases. This will also help us develop more patient-specific targeted therapies in a timely fashion. I am pretty excited about this new phase of technological development but will still use a cautious approach in choosing the right technique to carry out drug discovery. 

DT: How will the sector have to adapt? Are we likely to see a reduction in staffing levels?

RK: The pharma and biotech industry needs to have an open mindset for new tools that are being discovered and hire expertise in these areas to help them accelerate discovery. It should not be just competition between companies but opening up to a more collaborative approach to solve the challenges of medicine that humankind faces. Tools may become available, but not one person or company can solve all the medical challenges we face. Staffing levels might change with the advent of these new tools – hiring the right expertise to solve unmet medical needs should be higher on the priority list. Although AI can reach a point where it can predict the right outcomes for a patient, we still need a human component in the process to make the right, ethical and humane decision for the people who suffer from debilitating diseases.  

I am eagerly looking forward to the disruptive technologies that will appear in the next five to 10 years that might solve a variety of problems facing humankind. 

Biography

Rajesh Kumar received his PhD in Physics/Biophysics from the Indian Institute of Science, Bangalore, India. He has held positions as a postdoctoral fellow at Cold Spring Harbor Laboratory, the faculty of biochemistry at the Albert Einstein College of Medicine, and Biogen. He joined Schrodinger in the fall of 2022 as a Sr Director of Structural biology and has focused on building the lab infrastructure, including establishing a ELN system and contributing to small molecule drug discovery efforts on proteins involved in inflammation and oncology.

From DDW Volume 26 – Issue 4, Fall 2025 – Read the digital issue