IGN: Invariable gene set-based normalization for chromatin accessibility profile data analysis

Abstract

Chromatin accessibility profiles generated using ATAC-seq or DNase-seq carry functional information of the regulatory genome that controls gene expression. Appropriate normalization of ATAC-seq and DNase-seq data is essential for accurate differential analysis when studying chromatin dynamics. Existing normalization methods usually assume the same distribution of genomic signals across samples; however, this assumption may not be appropriate when there are global changes in chromatin accessibility levels between experimental conditions/samples. We present IGN (Invariable Gene Normalization), a method for ATAC-seq and DNase-seq data normalization. IGN normalizes the promoter chromatin accessibility signals for a set of genes that are unchanged in expression, usually obtained from accompanying RNA-seq data, and extrapolating to normalize the genome-wide chromatin accessibility profile. We demonstrate the effectiveness of IGN in analyzing central memory CD8⁺ T cell activation, a system with anticipated global reprogramming of chromatin and gene expression, and show that IGN outperforms existing methods. As the first chromatin accessibility normalization method that accounts for global differences, IGN can be widely applied to differential ATAC-seq and DNase-seq analysis. The package and source code are available on GitHub at https://github.com/zang-lab/IGN.