GLM relates genomic feature to spectral PC1

Generalized linear model built within each window, using PC1 of mean spectra as the outcome and each genomic feature (e.g. GC content) as the predictor.

Source