This repository contains the scripts used to produce the results and figures of the manuscript entitled "Biological Constraints Can Improve Prediction in Precision Oncology"

1. Bladder: the scripts used for data collection and training agnostic and mechanistic models to predict bladder cancer progression in non-muscle invasive bladder cancer patients.

2. Breast: the scripts used to collect the gene expression datasets and also to train the models to predict the response to neoadjuvant chemotherapy in patients with triple-negative breast cancer.

3. Prostate: the scripts used for collecting and processing the prostate gene expression datasets and for training the models to predict metastasis in primary prostate cancer.

4. CrossMechanism: the scripts used to assess the performance of different mechanisms in different prediction tasks.

5. GenePairs: contains the lists of mechanistic gene pairs: a) FFLs: feed-forward loops consisting of TFs and miRNA target genes; b) MYC_Pairs: pairs of genes up- and down-regulated by c-MYC; c) NOTCH_Pairs: pairs of genes up- and down-regulated by NOTCH signaling; d) metastasis_pairs: pairs of genes involved in cell adhesion, activation, and O2 response; e) Alzheimer_pairs: pairs of genes up- and down-regulated in the endothelial cells of patients with Alzheimer disease; f) diabetes_pairs: pairs of up- and down-regulated genes in the peripheral blood monocytes of patients with diabetes; and g) infection_pairs: pairs of up- and down-regulated genes in the immune cells with viral infections.