Pranav M. Khade
I know a little about proteins and how to design them
About Me
I have collaborated with cheminformatics professors, mathematicians, statisticians, experimental scientists, AI experts, and even civil engineers. I have been a speaker at a few prestigious conferences. I love to work in a diverse and open workplace where creativity, diversity, and innovation are valued.
I currently work as a Scientist at Gilead Sciences, and my research duties involve computational design of therapeutic proteins, including Antibodies, VHHs and immunization targets using AI ML methods I developed and industry standards. I have completed three degrees in life sciences including a Ph.D. degree in Computational Biology. Master’s degree in Bioinformatics and a Bachelor’s degree in Biotechnology and an industry postdoc at Prescient Design (Genentech Inc.).
During my Ph.D. I have successfully conceptualized, initiated, designed, and implemented several novel and innovative ideas, carrying them out to successful completion that resulted in seven publications. I received the Research Excellence Award and graduation honor for it. I have also built a software package called PACKing and Motion Analysis (PACKMAN) around my research, which has tens of thousands of downloads on the Python Package Index (PyPI) (https://github.com/Pranavkhade/PACKMAN). The ultimate goal of my thesis has been to explain the global motions of the proteins using protein packing information, and I have made significant progress with this. Along with the research work, I have also completed various courses on diverse topics such as Structural Bioinformatics, Statistical Bioinformatics, Bioinformatics Algorithms, Systems Biology, and many more.
During my Master’s training, I learned more about Bioinformatics in Cell Biology, Immunology, Structural Biology, Genetics, Omics, Evolution, Cheminformatics, Data Mining, and Biostatistics. Overall, my Masters’s degree included a detailed overview of Bioinformatics. I learned everything from predicting epitomes given viral protein sequences to relating Single Nucleotide Polymorphism to phenotypes. My Masters’s research project was in Cheminformatics, where I designed a method to overcome the shortcomings of stochastic conformer generators.
Here are a few technologies I've been working with recently:- Graph Neural Networks (GNNs)
- Attention Mechanisms
- Explainable Artificial (XAI)
- Protein Specific AI Architectures
- Delaunay Tessellations
- Protein Geometry
- Elastic Network Models
Experience
- Working on computational design of therapeutic proteins, including Antibodies, VHHs and immunization targets using AI ML methods I developed and industry standards.
- Collaborating with experimental scientists to validate the designs.
- Leading a team of researchers in the development of novel protein engineering methods.
- Participate in cutting-edge development of AI/ML models for drug discovery and scientific applications.
- Development and optimize exploratory analysis tools for interpreting multi-dimension dataset generated during biologics discovery workflow.
- Apply and advance artificial intelligence and machine learning (AI/ML) approaches, particular generative AI models, across stages of large molecule workflows, including molecular property predictions, AI-guided candidate selection and engineering.
- Work closely with data engineers to manage and preprocess large datasets, ensuring data quality and pipeline scalability.
- Communicate findings and insights to stakeholders through reports, presentations, and visualizations.
- Engage actively in the evaluation and coordination of collaborations with academic institutions, startups, and outsourcing partners.
- Stay updated on the latest advancements in AI/ML and computational biology and apply new techniques to improve existing workflows.
- Mentor and provide guidance to junior data scientists and research associates.
- Worked with group of machine learning expert in Prescient Design division.
- Built a novel Graph Neural Network to predict developability
- Work on Ab and TCR binding prediction with AI ML and Novel Coarse-grained MD
- Work on investigating geometric patterns responsible for molecular interactions.
- Got invited to talk about my research at PEGS 2024 (Boston)
- Aided in research input into four grants from which $1M NSF (1856477) was awarded in 2019 and several others under review
- Served on Bioinformatics and Computational Biology Graduate Student Organization Committee (2019)
- API, Library and GUI development experience.
- Designed a public library for python around my research and it has more than 60k downloads.
- Continous integration (CI) experience with Travis CI and GitHub Actions.
- Git repo maintainance experience.
- Sphinx documentation and Readthedocs implementation.
Education
August 2018 - April 2022
Ph.D. in Computational Biology
Iowa State University, Ames, IA
- Duration: 4 years 8 months (including 1 year rotations)
- Total publications: 8 (4 first author)
Extra curricular:
- Served on Bioinformatics and Computational Biology Graduate Student Organization Committee (2019)
2014 - 2016
Masters in Bioinformatics
Bioinformatics Centre, University of Pune
GPA:
5.34 out of 6
- Thesis Project: Systematic Conformer Generation (Cheminformatics)
- Training: Systems Biology, Immunology, Structural Biology, Genetics, Omics, Evolution, Cheminformatics, and Data Mining
Extra curricular:
- Won a silver medal as a captain of the volleyball team at University.
2011 - 2014
B.Sc in Biotechnology
Abasaheb Garware College, University of Pune
GPA:
First Class
- This degree equips me with experience communicating with and thinking like wet lab scientists.
- Designed and carried out wet lab experiments firsthand.
- Formal training in Animal and Plant tissue culture, microbiology, fermentation, Chromatography and many more wet lab techniques.
Projects
GNN
Protein-Protein Interactions
Protein Engineering
Therapeutic Applications
BRIDGE Encoder
Biophysical Representation of Interfaces via Delaunay-based Graph Embeddings (BRIDGE), a coarse-grained graph neural network that captures embeddings containing meaningful information about protein-protein interactions. The BRIDGE model takes as input graphs defined by the Delaunay tesselation of the individual chains and is pre-trained to predict the Delaunay adjacency at the protein-protein interface. The model achieves state-of-the-art performance in this task. The biophysical information captured by the BRIDGE embedding layer due to this pre-training task can further be used for downstream tasks, including for therapeutically relevant property prediction.
GNN
Antibody Developability
Protein Engineering
Delaunay Tessellations
Delaunay Graph Neural Network (D-GNN)
Meaningful Biological Priors as Guiding Constraints for Graph Neural Network-Based Antibody Developability Prediction.
Protein Dynamics
Protein Engineering
hdANM
The motions predicted with this new elastic network model provide important conceptual advantages for understanding the underlying biological mechanisms. As a matter of fact, the generated hinge movements are found to resemble the expected mechanisms required for the biological functions of diverse proteins. Another advantage of this model is that the domain-level coarse graining makes it significantly more computationally efficient, enabling the generation of hinge motions within even the largest molecular assemblies, such as those from cryo-electron microscopy.
Protein Hinges
Protein Dynamics
Graph Theory
Delaunay Tessellations
Protein Hinge Prediction
This was a fountation study before hdANM. This new method to identify hinges within protein structures is called PACKMAN. The predicted hinges are validated by using permutation tests on B-factors. Hinge prediction results are compared against lists of manually curated hinge residues, and the results suggest that PACKMAN is robust enough to reproduce the known conformational changes and is able to predict hinge regions equally well from either the open or the closed forms of a protein.
Protein Entropy
Protein Dynamics
Voronoi Tessellations
Protein Entropy Estimation
Even though the method is simple, the entropies computed exhibit an extremely high correlation with the entropies previously derived by other methods based on quasi-harmonic motions, quantum mechanics, and molecular dynamics simulations.
Protein Compliance
Elastic Network Models
Protein Structural Compliance
This present study considers only elastic network models, but the approach could be applied further to conventional atomic molecular dynamics. Compliance is found to have a slightly better agreement with the experimental B-factors, perhaps reflecting its bias toward the effects of local perturbations, in contrast to mean square fluctuations.
Protein Fluctuations
Elastic Network Models
Protein Fluctuations
This study concludes that It follows that protein fluctuations should be considered not just as the intrinsic fluctuations of the internal dynamics, but also equally well as responses to external solvent forces, or as a combination of both.
Achievements
NSF Travel Grant
Support to attend “International Conference on Mathematical Multiscale Modeling in Biology”- Guanacaste, Costa Rica.
IGIB-GNR Scholarship
Issued for Excellent performance in the entrance and academics at the Bioinformatics Centre, University of Pune.
DBT Fellowship
For each semester of M. Sc., top-performing students are awarded a monthly fellowship.
BCB Travel Fund
Based on my performance and work presentation
Get in Touch
My inbox is always open. Whether you have a question or just want to say hi, I’ll try my best to get back to you!