Welcome to the discussion group for users and developers of the Irreproducible Discovery Rate (IDR) statistical framework used to threshold and analyze reproducibility of quantitative biological experiments such as ChIP-seq data.

IDR was used extensively in the analysis of ChIP-seq data from the ENCODE (Encyclopedia of DNA elements) consortium. It is part of the official ENCODE/modENCODE data standards.

The statistcal method was developed by Qunhua Li, Ben Brown, Haiyan Huang and Peter Bickel at U.C. Berkeley. See http://projecteuclid.org/euclid.aoas/1318514284 for the publication in Annals of Applied Statistics (Li et al. 2011) introducing the statistical framework. For an intuitive description of the IDR method and a description of the various plots generated by the IDR package read this IDR-101 document compiled by Qunhua Li.

The method was further adapted, tweaked and deployed for systematic analysis of ChIP-seq data by Anshul Kundaje and others as part of the ENCODE consortium.

A manuscript detailing practical aspects of the IDR pipeline is in the works. A beta version of the pipeline is available at https://sites.google.com/site/anshulkundaje/projects/idr

Integrating IDR into NGS pipelines

Joseph Powell