ComBat-Seq: batch effect adjustment for RNA-Seq count data

Abstract

The benefit of integrating batches of genomic data to increase statistical power in differential expression is often hindered by batch effects, or unwanted variation in data caused by differences in technical factors across batches. It is therefore critical to effectively address batch effects in genomic data. Many existing methods for batch effect adjustment assume continuous, bellshaped Gaussian distributions for data. However in RNA-Seq studies where data are skewed, over-dispersed counts, this assumption is not appropriate and may lead to erroneous results. Negative binomial regression models have been used to better capture the properties of counts. We developed a batch correction method, ComBat-Seq, using negative binomial regression. ComBat-Seq retains the integer nature of count data in RNA-Seq studies, making the batch adjusted data 1 compatible with common differential expression software packages that require integer counts. We show in realistic simulations that the ComBat-Seq adjusted data result in better statistical power and control of false positives in differential expression, compared to data adjusted by the other available methods. We further demonstrated in a real data example where ComBat-Seq successfully removes batch effects and recovers the biological signal in the data.

Set-Wise Differential Interaction between Copy Number Alterations and Gene Expressions of Lower-Grade Glioma Reveals Prognosis-Associated Pathways

“…The RNA sequencing (RNA-seq) data has 60,483 transcripts in total, and 13,339 transcripts were mapped to the ensemble identifiers of all the pathway information in the 1335 pathways. Since the RNA-seq data had different batches, a batch effect adjustment was performed with Combat-seq program [ 29 ]. After the adjustment, the RNA-seq data were normalized using the quantile normalization method.…”

Rearing medium dictates variability across replicates in untreated and arsenic challenged zebrafish larvae

“…RNAseq analysis was done on the untreated samples from two independent datasets where larvae from one sample set were reared in egg water (Gene Expression Omnibus (GEO) accession number GSE151291) and the other was reared in embryo medium (GEO accession number GSE156420). To adjust the batch effect, we adopted a method designed for RNA-seq count data implemented by Combat-seq [48]. The method keeps the negative binomial distribution of RNA-seq reads count and the integer nature of the data.…”

Comparing tuberculosis gene signatures in malnourished individuals using the TBSignatureProfiler

“…We analyzed RNA-seq data from enrollment PaxGene tubes from a subset of 23 severely malnourished individuals with TB and 15 severely malnourished tuberculin skin test positive (TST ≥5 mm) household contacts as previously described [19]. The data were batch corrected using ComBat-Seq [47,48] ( Supplementary Figure 1). Differential expression between TB and LTBI samples produced 6706 differentially expressed features using an adjusted p-value (FDR) cutoff of 0.01, including 4913 protein coding genes, 1052 lncRNAs, 135 T cell receptive elements, 19 immunoglobulin genes, and 13 miR-NAs.…”