search

neurogenomics / ConsensusPeak

GNU General Public License v3.0
0 stars 0 forks source link

Add replicate analysis for broad peaks #4

Open Tomrrr1 opened 8 months ago

Tomrrr1 commented 8 months ago

IDR thresholding does not work on broad marks (it should be ok for histones that produce narrow marks, like H3K27ac: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7808876/#:~:text=According%20to%20the%20ENCODE%20guidelines,H3K4me3%2C%20H3K9ac%2C%20and%20H3K27ac).

We can use a simple overlap to call consensus peaks for broad marks. The function findOverlapsofPeaks from the ChIPpeakAnno package seems useful. The function returns a data frame of consensus peaks. We can specify the minimum overlap required for a peak to be considered a consensus using the minoverlap option. One of the drawbacks of this function is that it can only handle up to four replicates.

Tomrrr1 commented 8 months ago

Would be worth trying the rmspc package. I only considered MSPC for narrow marks, but nothing in the documentation suggests the method would not be suitable for broad marks. Regardless, I will still include the simple overlap function as I think it is nice to have.

Tomrrr1 commented 8 months ago

One of the drawbacks of findOverlapsOfPeaks is that it can only handle <5 replicates. Need to think of alternatives that allow for more replicates.