HMS Logo    BWH Logo    BWH Logo

Biological sequence models (theory, design, and application) Postdoctoral Fellow

Opening for a Postdoctoral Research Fellow to join the Gibson Lab at Harvard Medical School (HMS), Brigham and Women’s Hospital (BWH), and the Broad Institute of MIT and Harvard. We leverage tools from machine learning (ML) and control theory to better understand biological systems. While specific research questions may evolve over time, a key hallmark of our work is the careful co-design of computational and experimental models. Biological systems are simply too complex to be probed through either paradigm alone.

This specific project is focused on the application and development of (long context) biological sequence models. We are particularly interested in models that use state space models as their back bone (e.g. Evo is based on the Hyena model). You will also explore how embeddings from these models can be used to improve the performance of other models designed in the lab that learn dynamics from time series data. One question we have: can embeddings from DNA sequence models be used as priors for species interactions or be used to transfer learned species interactions from one experiment to new species in a different experiment. For this work you will have access to some unique computing resources we have on premises in our department (32 A100s, 40 H200s, and more planned). We also have the capacity to perform in vitro or in vivo validation of model predictions.

Qualifications

About the lab and fellow appointment

The Gibson Lab is physically located in the Division of Computational Pathology at BWH. All fellows have a triple appointment at BWH, HMS and the Broad. If necessary to conduct you work an official appointment at MIT can also be secured. For more information please go to the lab website: https://gibsonlab.io

Applications Process

Submit: (1) cover letter or research statement; (2) curriculum vitae; (3) two most relevant publications; (4) names and contact information of three individuals who can serve as references to: Travis Gibson, tegibson@bwh.harvard.edu. If you wish to chat briefly over Zoom before submitting materials to learn more details about our ongoing work, please inquire about this possibility