StockMeth: a computational framework to predict Noisy regions from ChIP experiments
Chromatin immunoprecipitation (ChIP) followed by sequencing is a standard technology in biological laboratories providing genome-wide information related to the interaction of proteins with DNA sequences. Recent studies have reported some limitations, in respect of the reliability, for ChIP experiments especially when the protein interact at specific genomic loci, such as in proximity to highly transcribed genes and particular euchromatin regions. There is an increasing evidence in the scientific community that specific genomic regions are highly reactive to proteins in an unspecific manner raising the question to what extent a genomic locus with significant peaks for a given protein is reliable for downstream interpretative analysis. For this, we have developed a computational method able to score DNA regions based on the reproducibility of protein binding sites and classified these regions according to the implemented reproducibility score. Furthermore, integration of methylation values at the classified regions, allowed to predict noisy regions with good sensitivity.
Keywords: Data Mining, Computational Biology