We offer two machine learning approaches as poosible downstream applications for TEPIC gene-TF scores:

INVOKE is a based on linear regression to identify key regulators in a sample of interest. All genes are treated as they are regulated similiary, thus TFs that affect the expression of many genes in the considered sample can be revealed.

DYNAMITE uses logistic regression to identify TFs that can classify genes as being up or downregulated between tissues/samples. This can be used to find TFs that might be related to differences between two distinct cell types or cell states.

EPIC-DREM links TFs to groups of genes exhibiting similar expression changes observed in time-series data. It uses TEPIC to compute time-point specific TF binding predictions from temporal epigenomic data sets.

For details on the methods, please check the individual subfolders as well as the documentation.