In this study, we present two applications of the GRridge method on count-based (mi)RNAseq, data that were generated from difficult impure samples, i.e. self-collected cervico-vaginal specimens and blood-platelets. We hypothesize the use of co-data can improve the performance of a classification model. We show the usage of different types of co-data from internal and external sources.
We also present additional functionalities of the method including better stability of the solutions, penalty estimation for overlapping groups like gene signature pathways, automatic selection of relevant co-data sources and post-hoc feature selection by adding L1-penalty.
Full reference: Novianti PW, Snoek B, Wilting SM, Van de Wiel MA (2017). Better diagnostic signatures from RNAseq data through use of auxiliary co-data. Bioinformatics, 33, 1572-1574.