Batch effect correction methods for NASA GeneLab transcriptomic datasets

Introduction: RNA sequencing (RNA-seq) data from space biology experiments promise to yield invaluable insights into the effects of spaceflight on terrestrial biology. However, sample numbers each study are low due limited crew availability, hardware, and space. To increase statistical power, RNA-seq datasets different missions often aggregated together. this can introduce technical variation or “batch effects”, differences in handling, processing, platforms. Several computational methods have been developed correct for batch effects, thereby reducing their impact true biological signals. Methods: In study, we combined 7 mouse liver NASA GeneLab (part Open Science Data Repository) evaluate several common effect correction (ComBat ComBat-seq sva R package, Median Polish, Empirical Bayes, ANOVA MBatch package). Principal component analysis (PCA) was used identify library preparation method mission as primary sources among variables dataset. We next quantitatively evaluated ability indicated identified variable using following criteria: BatchQC, PCA, dispersion separability criterion, log fold change correlation, differential gene expression analysis. Each variable/correction combination then assessed a custom scoring approach optimal dataset, by geometrically probing all allowable functions an aggregate volume-based measure. Results Discussion: Using described dataset variable/ComBat pair out ranked other candidate pairs, suggesting that should be corrected ComBat prior downstream describe multi-study visualization portal which will allow users access publicly available ‘omics data, select multiple studies combine analysis, examine presence absence metrics. If user chooses perform correction, here implemented use specific