Mayank Vats

Biologics CMC · Downstream Process Development

Molecular modeling · ML for process development

As part of the Biologics CMC organization, I sit at the interface of Drug Discovery and Clinical Trials. I specialize in downstream process development to accelerate “Speed-to-Clinic”, with work spanning the entire development lifecycle, from bench-scale unit operation optimization and tech transfer to advanced digital workflows. The digital side includes leveraging molecular modeling for in silico biophysical investigations, mechanistic modeling of chromatography, and developing machine-learning approaches for rapid process development across traditional monoclonal antibodies and newer modalities like fusion proteins and antibody-drug conjugates (ADCs).

My background is rooted in bridging molecular mechanism and physical process. During my PhD in Chemical and Biological Engineering at Rensselaer Polytechnic Institute, I used molecular dynamics simulations to elucidate how multimodal ligand surfaces govern protein retention in chromatography. Today, I work deliberately at the boundary of data science and the bench. I adapt machine-learning methods such as QSAR for retention prediction and neural-network featurization of protein structures to solve core challenges owned by the process development community.

latest work

Accelerated Ultrafiltration/Diafiltration Process Development to Expedite Early-Stage High-Concentration Biologics

ACS BIOT 2026

Atlanta, GA, USA · March 2026

Implementation of Multimodal Anion Exchange Chromatography to Address Product Quality Challenges and Downstream Platform Limitations: A Case Study

Julie Robinson, Mayank Vats, and Michael Hartmann

Journal of Chromatography A, Apr 2025

Abs DOI

Flow through anion exchange chromatography (AEX) has provided reliable process-and product-related impurities removal as well as viral clearance for monoclonal antibodies (mAbs). The application of AEX to molecules with more complex impurity profiles or non-platform characteristics such as a low pI can become challenging because viral clearance considerations often constrain the AEX step design space. Multimodal anion exchange chromatography (MMAEX) can address the limitations of the platform AEX step while still allowing a “platform-like” manufacturing process. This work presents a case study on polishing step development for a Fc-fusion protein with pI < 6.5, high surface hydrophobicity, and aggregate content of up to 35% in the harvested cell culture fluid. An integrated computational and experimental high throughput screening (HTS) workflow was implemented to rapidly identify the MMAEX resin Capto Adhere at pH 5.5 as a viable alternative to the conventional AEX flow through polishing step. This work presents the impact of pH, conductivity, load, and load impurity level/composition on MMAEX step performance. Step performance was measured based on yield, HMW clearance, residual HCP and DNA clearance, as well as viral clearance. HMW clearance varied based on the starting load HMW level (from 5 -35%) and the highest relative clearance was obtained at a moderate HMW load level. Residual HCP and DNA showed a strong dependance on load HMW level, highlighting the competitive adsorption behavior that can impact resin-protein interactions in complex mixtures. Both residual HCP and DNA were removed to below quantification at pH 5.5. Viral clearance of up to 4 logs of XMULV and MVM was demonstrated at pH 5.5 using process relevant feed streams. The workflow presented here demonstrates how the integration of in silico modeling and high throughput screening (HTS) can streamline process development and enable rapid polishing step optimization. These results also underscore the impact of feed impurity and impurity composition on MMAEX resin performance.