TANDEM is a two-stage penalized linear regression approach to that uses upstream (genomics) and downstream (gene expression) to predict drug response. It results in models that are more interpretable while maintaining similar predictive performance.
- In the first stage, TANDEM explains response using upstream features (mutations, copy number, methylation and cancer type)
- In the second stage, TANDEM explains the remainder using downstream features (gene expression).
Publications related to this work
- TANDEM: a Two-stage Approach to Maximize Interpretability of Drug Response Models Based on Multiple Molecular Data Types, Aben et all, Bioinformatics 2016 (ECCB proceedings) (link)
The R package is available on CRAN.