AndreKim/cancer-mutation-classifier

Cancer Mutation Classifier

Classifies a genetic mutation into one of 9 clinical categories (driver vs. passenger) from its gene, amino-acid variation, and supporting clinical / scientific evidence text.

Baseline model: TF-IDF (text) + one-hot (Gene, Variation) features, calibrated LinearSVC, log-loss ~1.00 on the MSKCC validation split.

Full code, methodology, and the 5 data-leakage bugs found and fixed in the original pipeline: see the GitHub repository (update this link once pushed).

Portfolio / research project -- not a medical device, not for clinical use.