MACS2 bdgdiff output

[blancha]

Hi,

The last column in the output from MACS2 bdgdiff is always -1, and not the log10 likelihood ratio as specified in the online documentation.

https://github.com/taoliu/MACS/wiki/Call-differential-binding-events

I've tried with the latest version available from GitHub, but I still always get -1 in the last column. The peaks called do appear to be correct, however.

Does anyone have a solution or an alternative software I could use?
I'm not getting any answers from the MACS2 mailing list.
I tried DiffBind, but I don't have any replicates unfortunately, so it doesn't work.
I have 2 treatment conditions (IP) with their controls (input).
I want to do (treament 1 - input 1) vs (treatment2 - input2).
The output from MACS2 bdgdiff would be correct were it not for all the -1s.

Here is a sample output from MACS2 bdgdiff.

track type=narrowPeak name="peak" description="peak" nextItemButton=on
chr1 16861631 16861989 ../../bdgdiff_github/IP_IC_vs_IP_IC_32/IP_IC_vs_IP_IC_32_encodePeak1 0 . 0.00000 0.00000 0.00000 -1
chr1 16872179 16872757 ../../bdgdiff_github/IP_IC_vs_IP_IC_32/IP_IC_vs_IP_IC_32_encodePeak2 0 . 0.00000 0.00000 0.00000 -1
chr1 16873932 16874461 ../../bdgdiff_github/IP_IC_vs_IP_IC_32/IP_IC_vs_IP_IC_32_encodePeak3 0 . 0.00000 0.00000 0.00000 -1

Thank you for your help,

Alexis

[TiborNagy]

The 5th, 6th, 7th columns are 0, so it can not be larger than -1. I think you made some minor error in the command line.

[blancha]

Thanks for your answer.

What should be in the columns instead of 0s?

Here is the command I used.
I can't find the mistake.

----

/usr/local/tools/MACS-2_github_27_03_2014/bin/macs2 bdgdiff \
--t1 ../../macs2/IP_IC_vs_input_IC/IP_IC_vs_input_IC_treat_pileup.bdg \
--c1 ../../macs2/IP_IC_vs_input_IC/IP_IC_vs_input_IC_control_lambda.bdg \
--t2 ../../macs2/IP_IC_32_vs_input_IC_32/IP_IC_32_vs_input_IC_32_treat_pileup.bdg \
--c2 ../../macs2/IP_IC_32_vs_input_IC_32/IP_IC_32_vs_input_IC_32_control_lambda.bdg \
--depth1 30080647 --depth2 26032880 \
--o-prefix ../../bdgdiff_github/IP_IC_vs_IP_IC_32/IP_IC_vs_IP_IC_32 \
1> bdgdiff_IP_vs_IP_32.sh_output 2> bdgdiff_IP_vs_IP_32.sh_error

----

-bash-4.1$ more bdgdiff_IP_vs_IP_32.sh_error
INFO @ Mon, 31 Mar 2014 10:09:30: Read and build treatment 1 bedGraph...
INFO @ Mon, 31 Mar 2014 10:11:40: Read and build control 1 bedGraph...
INFO @ Mon, 31 Mar 2014 10:14:45: Read and build treatment 2 bedGraph...
INFO @ Mon, 31 Mar 2014 10:18:04: Read and build control 2 bedGraph...
INFO @ Mon, 31 Mar 2014 10:28:13: Write peaks...
INFO @ Mon, 31 Mar 2014 10:28:13: Done

-----

-bash-4.1$ more ../../macs2/IP_IC_vs_input_IC/IP_IC_vs_input_IC_treat_pileup.bdg
chrY 0 10252 0.00000
chrY 10252 10481 1.00000
chrY 10481 10623 0.00000
chrY 10623 10649 1.00000

-----

-bash-4.1$ more ../../macs2/IP_IC_vs_input_IC/IP_IC_vs_input_IC_control_lambda.bdg
chrY 0 11523 2.23098
chrY 11523 11555 2.55359
chrY 11555 14172 2.23098
chrY 14172 14286 2.55359
chrY 14286 20562 2.23098

[blancha]

Problem resolved with updated MACS2

Hi,

Well, it turned out to be entirely my fault, unsurprisingly .
Tao Lio helped me figure it out on the MACS2 mailing list.

I wasn't updating the PYTHONPATH environment variable, so I was still using the older MACS2 2.0.10.20120913 version, even though I thought I had installed the latest version.
With MACS2 2.0.10.20131216 , I do get the log 10 likelihood ratio in the last column.

Thank you for your help.

-bash-4.1$ more IP_IC_vs_IP_IC_32_c3.0_cond1.bed
track name="condition 1 (peaks)" description="unique regions in condition 1" visibility=1
gi|555853|gb|U13369.1|HSU13369 98 2298 ../../bdgdiff_humRibosomal/IP_IC_vs_IP_IC_32/IP_IC_vs_IP_IC_32_cond1_1 15.38272
gi|555853|gb|U13369.1|HSU13369 2652 2919 ../../bdgdiff_humRibosomal/IP_IC_vs_IP_IC_32/IP_IC_vs_IP_IC_32_cond1_2 15.81923

Best,

Alexis