OMICS INTEGRATION FOR PRECLINICAL GEMMA DATA

OMICS INTEGRATION FOR PRECLINICAL GEMMA DATA

We developed a computational pipeline with accompanying scripts for the multi-omics integration of data from GEMMA’s preclinical FMT mouse experiments. We apply machine learning to find the most statistically relevant and discriminant features associating with ASD as well as with FMT dietary groups beginning at the individual data type level. The top features are tested with logistic regression to assess their effect sizes and directionality. Multiomics Integration is additionally performed to detect potential interactions between microbiome, genetics and molecular biomarkers using MixOmics and correlation based testing. The pipeline and approach should be applicable for other data types and designs with minimal changes.