macs

## meaning of pileup value from macs2 (https://groups.google.com/forum/#!topic/macs-announcement/c__KCotHsok)

Dear Fred,

Now, what does that pileup value really mean? 

ChIP pileup values are calculated in this way: 1) Sequencing reads are extended into N bps fragments according to fragment size prediction step; 2) For each position, MACS2 computes how many fragments can be found; 3) Only if ChIP has more reads than control, these values will be multiplied by control_depth/ChIP_depth (default behavior). The 'pileup' value you saw in xls file is the value at peak summit, which means how many fragments you can find covering the peak summit. 

Can I use the pileup value from experiment 1 to compare results with experiment 2?

You can't directly use them if experiments have different sequencing depths. You can use '--SPMR' option for 'callpeak' command to get normalized pileup values in million reads. Then these values can be compared.
 
 Does it directly correlate with the number of tags on each peak summit?

 Yes. 

 In order to filter out weak peaks using pileup value, you can draw a distribution of pileup value at peak summit for all peaks, then select a reasonable cutoff. Or you can consider both p or q-values and fold-enrichment values since these measurements take control sample in consideration. 

 Best,
 Tao