Position Description
Location: Toronto, Ontario, Canada (Split between The Hospital for Sick Children and the Structural Genomics Consortium).
Term: Multi-year project (Estimated 2026-2028).
Supervision: Dr. Michal Koziarski (SickKids/UofT) and Dr. Matthieu Schapira (SGC/UofT).
Industry partnership: A Senior Scientific Director, Computational Drug Discovery from a large pharmaceutical company agreed to act as Industry Mentor.
We are seeking a highly motivated Post-Doctoral Fellow to lead a cutting-edge project focused on advancing machine learning (ML) methodologies for the accurate prediction of small-molecule binding affinity. This project addresses the fundamental challenges of data scarcity and generalization in drug discovery by leveraging large-scale datasets from DNA-encoded library (DEL) and affinity selection mass spectrometry (ASMS) platforms.
As a Mitacs Target 2035 post-doctoral fellow, you will work in an interdisciplinary team with academic, SGC, and industry scientists and benefit from training supported through Mitacs.
Core Objectives
You will be responsible for executing two tightly integrated research objectives:
- Multi-fidelity binding affinity prediction pipelines: develop efficient strategies for integrating existing foundation ML models with target-specific experimental screening data and computationally intensive physics-based methods.
- Novel foundation ML models trained on DEL/ASMS data: develop novel foundation ML models capable of learning transferable representations of molecular recognition directly from DEL and ASMS datasets spanning hundreds of proteins. These models will support generalizable predictions for unseen proteins and ligands.
Role Responsibilities
- Benchmarking: Systematically evaluate existing foundation models (e.g., Boltz-2, AQAffinity, Uni-Mol) for binding affinity prediction.
- Model Engineering: Develop novel fine-tuning strategies tailored to DEL/ASMS datasets.
- Pipeline Integration: Implement iterative workflows where ML predictions guide the selection of compounds for hit-to-lead optimization.
- Dissemination: Produce at least two peer-reviewed publications and assemble an open-source computational library integrating foundation ML models and multi-fidelity optimization components.
Qualifications
- PhD in Computational Chemistry, Computer Science, Bioinformatics, or a related field.
Technical Expertise: Strong background in representation learning, foundation models, and multi-fidelity optimization. - Domain Knowledge: Familiarity with physics-based molecular modeling (ex: FEP).\
- Skills: Experience with large-scale ML training, handling "noisy" experimental data (like DEL).
Open Science: This project operates within an open science framework, ensuring your work has a high impact on the global drug discovery community.
Additional Information
This position is based at The Hospital for Sick Children (SickKids) and has been created to advance research efforts aligned with the goals of Target 2035, a global open-science initiative focused on enabling the discovery of chemical probes for every human protein. Successful candidates will contribute to cutting-edge projects in protein science, data generation, and early-stage drug discovery supported by SickKids affiliated scientist that share the principles of open science.
Find more information and apply here: https://career.sickkids.ca:8001/psc/CRPRD/CAREER/HRMS/c/HRS_HRAM_FL.HRS…