Comparative ChIP-seq (Comp-ChIP-seq): a novel computational methodology for genome-wide analysis

Abstract

Chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-seq) is a pivotal technique for understanding the functionality of the chromatin-bound factors and for mapping the functional elements of the genome. In order to evaluate cell- and disease-specific changes in the interacting strength of chromatin targets, ChIP-seq signal across multiple conditions must undergo robust normalization. However, this is not possible using the standard ChIP-seq scheme, which lacks a reference for the control of biological and experimental variabilities. While several studies have recently proposed different solutions to circumvent this problem, substantial technical and analytical differences among methodologies could hamper the experimental reproducibility. Here we provide a practical binary decision-making process to experimentally implement a normalizing method for comparative ChIP-seq across different samples. In addition, we evaluate side-by-side the current computational approaches for normalizing using a reference internal genome. Finally, we propose a local regression strategy to accurately normalize ChIP-seq data in a genome-wide manner. Overall, our proposed experimental and computational standard for comparative ChIP-seq (Comp-ChIP-seq) will increase experimental reproducibility, thereby reducing this major confounding factor in interpreting ChIP-seq results.