An approach for prioritizing candidate genes from RNA-seq using preclinical cocaine self-administration datasets as a test case

Abstract RNA-sequencing (RNA-seq) technology has led to a surge of neuroscience research using animal models to probe the complex molecular mechanisms underlying brain function and behavior, including substance use disorders (SUDs). However, findings from rodent studies often fail to be translated into clinical treatments. Here, we developed a novel pipeline for narrowing candidate genes from preclinical studies by translational potential and demonstrated utility of this model in three RNA-seq studies of rodent self-administration. This pipeline uses evolutionary conservation and preferential expression of genes across brain tissues to prioritize candidate genes, increasing the translational utility of RNA-seq in model organisms. We found only 1 differentially-expressed gene (DEG) in any dataset after correcting for multiple testing (FDR < 0.05 or < 0.1), raising concerns about false positives and low statistical power that may impact these and other RNA-seq datasets. Thus, we demonstrate the utility of our prioritization pipeline using an uncorrected p-value. Curiously, even with an uncorrected p-value of 0.05 we saw low overlap of DEGs across the 3 selected datasets. Thus, we also advocate for improved RNA-seq data collection, statistical testing, and metadata reporting that will bolster the field’s ability to identify reliable candidate genes and compare results across studies.