TADShop: systematic benchmarking and identification of topologically associating domains
摘要
Topologically associating domains (TADs) are structural units of chromatin organization. Their definition and identification rely on computational analyses of chromosome conformation capture data. Here we systematically assessed and compared 43 TAD identification strategies, none of which excelled in all tests, often displaying conflicting results. To benefit from the strengths and overcome the weaknesses of individual tools, we developed a dynamic programming framework to integrate the results of top-performing methods into a consensus list of TADs (ConsensusTAD). ConsensusTAD outperformed individual tools, and it can be applied to integrate results from any set of current or future TAD callers. To facilitate TAD identification and benchmarking, we developed TADShop (https://tadshop.unil.ch/), a web service that enables the comparison and use of TAD callers through an intuitive graphic user interface. By facilitating rigorous benchmarking and adoption of best performing tools, TADShop provides a standardized platform to foster robust and reproducible analyses of chromatin organization.