To mock or not: a comprehensive comparison o f mock IP and DNA input for ChIP-seq

Chromatin immunoprec pitation (IP) followed by sequencing (ChIP-seq) is the gold standard to detect transcription-factor (TF) binding sites in the genome. Its success depends on appropriate controls removing systematic biases. The predominantly used controls, i.e. DNA input, correct for uneven sonication, but not for nonspecific interactions of the IP antibody. Another type of controls, 'mock' IP, corrects for both of the issues, but is not widely used because it is considered susceptible to technical noise. The tradeoff between the two control types has not been investigated systematically. Therefore, we generated comparable DNA input and mock IP experiments. Because mock IPs contain only nonspecific interactions, the sites predicted from them using DNA input indicate the spurious-site abundance. This abundance is highly correlated with the `genomic activity' (e.g. chromatin openness). In particular, compared to cell lines, complex samples such as whole organisms have more spurious sites-probably because they contain multiple cell types, resulting in more expressed genes and more open chromatin. Consequently, DNA input and mock IP controls performed similarly for cell lines, whereas for complex samples, mock IP substantially reduced the number of spurious sites. However, DNA input is still informative; thus, we developed a simple framework integrating both controls, improving binding site detection.

Automated summary

Chromatin immunoprecipitation followed by sequencing is the gold standard for detecting transcription-factor binding sites, with success dependent on appropriate controls; while DNA input corrects for uneven sonication, mock IP corrects for nonspecific interactions, especially in complex samples like whole organisms. The two control types perform similarly in cell lines, but mock IP substantially reduces the number of spurious sites in complex samples, ultimately improving binding site detection.