Hi,
I've got a questions regarding MACS1.4, which I already posted in the MACS announcement forum. But as I didn't get any replies I thought I post it here and see if anyone can help.
I'm working with highly repetitive gene regions that are not included in the current genome assemblies. I have a few ChIPseq files I need to align to these regions and to do so I added the repetitive sequences as additional chromosome to the genome assembly and used this custom genome for the alignment. The alignment output is fairly similar to the output I get when aligning it to the normal genome, however when I call peaks with MACS1.4.2 something weird is happening. MACS does call a similar amount of peaks, however the peaks are in different locations compared to the normal genome files and the FDR values are worse, which I think is a consequence of estimating a different fragment length.
Here's the stderr:
normal genome
#1 tag size is determined as 50 bps
#1 tag size = 50
#1 total tags in treatment: 1826394
#1 total tags in control: 40575275
#1 finished!
#2 Build Peak Model...
#2 number of paired peaks: 20414
#2 finished!
#2 predicted fragment length is 337 bps
custom genome
#1 tag size is determined as 50 bps
#1 tag size = 50
#1 total tags in treatment: 2068421
#1 total tags in control: 40188963
#1 finished!
#2 Build Peak Model...
#2 number of paired peaks: 21543
#2 finished!
#2 predicted fragment length is 88 bps
The script for both analyses is exactly the same (with defined bandwidth of 250), the only difference is the genome file.
Any ideas what's going on and how to avoid the problem? I really would like to use only the custom genome in the future, but therefore I need to be sure that it does the same thing as the normal genome!
Thanks,
Jeannine
I've got a questions regarding MACS1.4, which I already posted in the MACS announcement forum. But as I didn't get any replies I thought I post it here and see if anyone can help.
I'm working with highly repetitive gene regions that are not included in the current genome assemblies. I have a few ChIPseq files I need to align to these regions and to do so I added the repetitive sequences as additional chromosome to the genome assembly and used this custom genome for the alignment. The alignment output is fairly similar to the output I get when aligning it to the normal genome, however when I call peaks with MACS1.4.2 something weird is happening. MACS does call a similar amount of peaks, however the peaks are in different locations compared to the normal genome files and the FDR values are worse, which I think is a consequence of estimating a different fragment length.
Here's the stderr:
normal genome
#1 tag size is determined as 50 bps
#1 tag size = 50
#1 total tags in treatment: 1826394
#1 total tags in control: 40575275
#1 finished!
#2 Build Peak Model...
#2 number of paired peaks: 20414
#2 finished!
#2 predicted fragment length is 337 bps
custom genome
#1 tag size is determined as 50 bps
#1 tag size = 50
#1 total tags in treatment: 2068421
#1 total tags in control: 40188963
#1 finished!
#2 Build Peak Model...
#2 number of paired peaks: 21543
#2 finished!
#2 predicted fragment length is 88 bps
The script for both analyses is exactly the same (with defined bandwidth of 250), the only difference is the genome file.
Any ideas what's going on and how to avoid the problem? I really would like to use only the custom genome in the future, but therefore I need to be sure that it does the same thing as the normal genome!
Thanks,
Jeannine
