ChIP-seq continues to assault bioinformatics teams with avalanches of data, but the upside of using this powerful approach brings some challenges along with it. Modeling background accurately requires a few tweaks, so before you get all excited about the double black diamond peaks that you’re seeing in regulatory motifs, you might want to factor in a few contributing sources of signal like Dr. Joel Rozowsky and his colleagues at Yale did in a recent Nature Biotechnology paper unveiling PeakSeq, an approach for scoring ChIP-seq results that improves accuracy.
Two factors PeakSeq takes into account when modeling ChIP-seq signals are:
- Tag Uniqueness: Since tags from some locations don’t always map to unique locations. Reads from these areas were filtered from the set.
- Chromatin Structure: The degree of openness of chromatin can influence the total amount of DNA that can be effectively analyzed in certain regions of the genome.
The team took these factors into consideration and chose to use input from the same cell line, under identical conditions as the experimental as a comparative control. PeakSeq was put to testing profiling Pol II and STATI binding on the Illumina Genome Analyzer. Read about all the details at: Nature Biotechnology, January 2009.