Fig. 4.

Identification of IBC-relevant Gene Signatures via Iterative Logistic Regression. A) Logistic regression was applied to each of the transcripts for GSE227597 (blue), GSE45581 (yellow), and GSE58477 (red). Transcripts were sorted based on their ability to resolve IBC and n-IBC samples. The intersection of the top 2000 transcripts were used to define the LR-specific IBC signature, S. B) LR signature fold-change between IBC vs. nIBC across three datasets. The bold-cases show when the fold-change value (mean (IBC)/mean (nIBC)) was statistically significant (p < 0.05) using a Students’ two-tailed t-test.