For 6 billion base pairs of the genomic DNA to be compacted into the nucleus of a human cell, DNA must tightly bend around a histone octamer to form the nucleosome, a compact packing unit. In this process, DNA bendability is a fundamental mechanical property that affects genome function. However, theoretical modeling is currently lacking to provide sequence-dependent quantification of intrinsic DNA bendability. In 2020, Basu and his coworkers developed a high-throughput “loop-seq” method to measure the bendability of tens of thousands unique DNA sequences in parallel (PMC7855230). Leveraging these valuable datasets, a team of researchers led by statistician Ji-Ping Wang and biologist Alec Wang at Northwestern NSF Simons Center for Quantitative Biology (CQuB) have developed a machine learning approach to accurately predict intrinsic DNA bendability of any DNA sequences (PMC8989542). To facilitate easy access, this tool, named DNAcycP, provides both a web server and a stand-alone Python package.
