search

jsh58 / Genrich

Detecting sites of genomic enrichment
MIT License
182 stars 27 forks source link

Appropriate area under the curve cut-off for ChIP-seq data #58

Closed HuynhNPT closed 4 years ago

HuynhNPT commented 4 years ago

I was wondering if it will be an issue if we're calling TF peaks instead of histone modification peaks? In MACS2, there's the option to specify whether we're calling narrow or broad peaks. Therefore, I was wondering if -a should be changed when processing TF or HM ChIP-seq. Please advice.

jsh58 commented 4 years ago

Thanks for the question. In MACS2, the --broad option simply lowers the statistical threshold to q< 0.1 and increases the maximum distance between significant sites by a factor of 4.

In Genrich, one can alter the peak-calling parameters: the statistical threshold (-p/-q) and the maximum distance (-g). You can also play with -a, although keep in mind that a short peak with high significance may have the same AUC as a large broad peak with low significance. Finally, you should consider using the -P argument to explore the effects of different peak-calling parameters more efficiently.

HuynhNPT commented 4 years ago

Thank you so much for your input. This is a good place for me to start optimizing.