Dear mikysyc2016, -l (--min-length) and -g (--max-gap) relate to the minimum peak size and the maximum distance between statistically significant peaks for the purposes of merging these into a single peak, respectively.
You can access the help pages by executing the following macs2 bdgdiff -h:
-l MINLEN, --min-len MINLEN
Minimum length of differential region. Try bigger
value to remove small regions. DEFAULT: 200
-g MAXGAP, --max-gap MAXGAP
Maximum gap to merge nearby differential regions.
Consider a wider gap for broad marks. Maximum gap
should be smaller than minimum length (-g). DEFAULT:
100
Hi Kevin,
If I use -l = fragment length i used as callpeaks and use -g as 100. i get 11294 peaks in common.bed and get 4692 peaks in con1.bed and get 2174 peaks in con2.bed. But I get 18347 peaks in con1 narrow.peaks and get 31410 peaks in con2 narrow.peaks. SO 11294+4692<18347 and 11294+2174<31410? do you think i need to change -g ?or I need to change peak length in narrow.peak as same as 200 then do diffpeak? Because the peak length is different after i use macs2 to do peak calling.
Miky
Have you checked some regions that you know should definitely be peaks? - how do the look? Yes, it is possible to refine your analysis by modifying the parameters but you should always have some sort of 'positive control' peaks that you know should exist (and negatives), which can help to guide you. In the past, I spent months going back and forth with different parameters for both MACS and HOMER, but nothing could ever get it exactly right (we tried every possible combination of values...).
Hi,
I use homer to do the process and also use its merge peak parameter. I use IGV to see the specific and overlap peaks, as you mentioned, I think some of the specific peak is real weak( some look like real), do not look like real peak for me. Do you have any suggestion about how to choose the real specifc peaks?
Thanks in advance.
Okay, there should be previous literature that states the binding regions for these TFs. You could overlap your regions with the previous identified regions, or indeed just check that the binding motifs at your identified peaks match the expected binding motifs for your TFs. HOMER provides some functions for motif analysis
Hi Kevin, If I use -l = fragment length i used as callpeaks and use -g as 100. i get 11294 peaks in common.bed and get 4692 peaks in con1.bed and get 2174 peaks in con2.bed. But I get 18347 peaks in con1 narrow.peaks and get 31410 peaks in con2 narrow.peaks. SO 11294+4692<18347 and 11294+2174<31410? do you think i need to change -g ?or I need to change peak length in narrow.peak as same as 200 then do diffpeak? Because the peak length is different after i use macs2 to do peak calling. Miky
Have you checked some regions that you know should definitely be peaks? - how do the look? Yes, it is possible to refine your analysis by modifying the parameters but you should always have some sort of 'positive control' peaks that you know should exist (and negatives), which can help to guide you. In the past, I spent months going back and forth with different parameters for both MACS and HOMER, but nothing could ever get it exactly right (we tried every possible combination of values...).
Hi, I use homer to do the process and also use its merge peak parameter. I use IGV to see the specific and overlap peaks, as you mentioned, I think some of the specific peak is real weak( some look like real), do not look like real peak for me. Do you have any suggestion about how to choose the real specifc peaks? Thanks in advance.
Well, what is your experiment? - it is a transcription factor or some other type of marker?
I did two TF ChIP-seq. And both of them express in same tissue. I want to t compare their binding sites( overlapped and specific).
Okay, there should be previous literature that states the binding regions for these TFs. You could overlap your regions with the previous identified regions, or indeed just check that the binding motifs at your identified peaks match the expected binding motifs for your TFs. HOMER provides some functions for motif analysis