teff: estimation of Treatment EFFects on transcriptomic data using causal random forest

Abstract

Abstract Motivation Causal inference on high-dimensional feature data can be used to find a profile of patients who will benefit the most from treatment rather than no treatment. However, there is a need for usable implementations for transcriptomic data. We developed teff that applies random causal forest on gene expression data to target individuals with high expected treatment effects. Results We extracted a profile of high benefit of treating psoriasis with brodalumab and observed that it was associated with higher T cell abundance in non-lesional skin at baseline and a lower response for etanercept in an independent study. Individual patient targeting with causal inference profiling can inform patients on choosing between treatments before the intervention begins. Availability and implementation teff is an R package available at https://teff-package.github.io. The data underlying this article are available in GEO, at https://www.ncbi.nlm.nih.gov/geo/ Supplementary information Supplementary data are available at Bioinformatics online.