This repository contains preliminary training data and a code stub for evaluating leave-multiple-chromosome-out cross-validation performance for the next version of TargetFinder (in development). This iteration uses a different formulation that fixes previous generalization problems, making the code more appropriate for predictive applications (in silico Hi-C).
Tested with Python 3.7, feather 0.4, numpy 1.15.4, pandas 0.23.4, sklearn 0.20.1, and xgboost 0.81.