Graph Transformer Research Probe for a Drug-Target Predictor
Overview
What this challenge is about.
You receive a public drug-target interaction dataset (around 50,000 drug-target pairs with labels and molecular graphs), a strong GIN baseline, and a starter GraphGPS implementation. Train both architectures with matched compute and report ROC-AUC, PR-AUC, and per-target-family performance. Run 3 ablations on GraphGPS (positional encoding, attention scope, head count). Write a publication-style 4-page memo on whether graph-transformers earn a research-quarter investment.
The Brief
What you'll do, and what you'll demonstrate.
Quantify whether graph-transformers beat strong message-passing GNNs on drug-target interaction prediction and recommend a research investment.
Earning criteria — what you'll demonstrate
- Implement and compare a Graph Transformer with a message-passing GNN
- Run controlled ablations on transformer hyperparameters
- Evaluate molecular learning models with per-family stratification
- Write publication-style research memos for leadership
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Research Scientist
Comparing graph-transformers against strong GNN baselines with rigorous ablations is exactly the day-one work of a research scientist on any graph-ML or chem-informatics team.
This challenge sharpens
- graph-transformers
- experiment-design
- drug-target-prediction
ML Researcher
Running compute-matched benchmarks and writing publication-style memos transfers directly to ML-researcher roles in industry research labs.
This challenge sharpens
- graph-neural-networks
- message-passing
- experiment-design
Applied AI Scientist
Translating a methodology result into a research-investment recommendation is core applied-AI-scientist work in pharma AI.
This challenge sharpens
- graph-transformers
- pytorch-geometric
- drug-target-prediction