Detection and correction of batch effect in transcriptomics analysis

With continuous development of sequencing technology, omics big data has become increasingly important for life science research. With accumulation of sequencing data and application of single-cell sequencing technologies, integration of massive transcriptomic datasets provides new opportunities and challenges in the post-genomic era. The batch effect introduced by spatial-temporal heterogeneity from different data sets and systematic errors from sequencing platform greatly affects effectiveness of transcript analysis and hinder discovery of true biological differences. This review discusses the causes and detection methods of batch effect in transcriptomic analysis. Mainstream correction models based on parametric and non-parametric estimation, integration algorithms for single-cell transcriptomes are summarized. Some practical recommendations for batch effect correction are provided, including suggestions for comprehensive transcriptomic analysis.