Rational design of peak calling parameters for TIP-seq based on pA-Tn5 insertion patterns improves predictive power

Epigenomic profiling provides insights into the regulatory mechanisms that govern gene expression. At a fundamental level, these mechanisms are determined by proteins that bind the DNA or modify the chromatin. Techniques such as ChIP-seq and CUT&Tag have been instrumental in mapping the binding sites of such proteins across the genome. Recent advances have led to the development of TIP-seq, a highly sensitive method devised to increase the number of unique reads per sample. Its design results in novel library features, which have not yet been explored with comparative analytics. Through the extensive assessment of bioinformatics tools and parameters we have developed an analysis pipeline that is ideally suited for TIP-seq data, including linear deduplication, read prioritisation and read shifting. Using transcription factor binding profiles (TFs), we show that our optimised pipeline greatly reduces the width of peaks to below 50% and more precisely aligns the peak summit with known motifs. A tutorial of the optimised peak calling is available on GitHub at https://github.com/neurogenomics/peak_calling_tutorial.git. Our methodological advancement substantially improves TIP-seq data quality, and the thoughtful design of analysis parameters is widely applicable to all pA-Tn5 based profiling assays.