Question: MACS returns an error

9 months ago by

jenyka_dm • 10 wrote:

Hi again! I am trying to run MACS on my treatment and control, but it comes back with an error. INFO @ Fri, 02 Mar 2018 16:36:30:

ARGUMENTS LIST:

name = MACS_in_Galaxy

format = BAM

ChIP-seq file = /galaxy-repl/main/files/023/877/dataset_23877240.dat

control file = /galaxy-repl/main/files/023/877/dataset_23877244.dat

effective genome size = 1.19e+08

tag size = 101

band width = 300

model fold = 32

pvalue cutoff = 1.00e-05

Ranges for calculating regional lambda are : peak_region,1000,5000,10000

INFO @ Fri, 02 Mar 2018 16:36:30: #1 read tag files... INFO @ Fri, 02 Mar 2018 16:36:30: #1 read treatment tags... INFO @ Fri, 02 Mar 2018 16:36:42: 1000000 INFO @ Fri, 02 Mar 2018 16:36:54: 2000000 INFO @ Fri, 02 Mar 2018 16:37:07: 3000000 INFO @ Fri, 02 Mar 2018 16:37:20: 4000000 INFO @ Fri, 02 Mar 2018 16:37:33: 5000000 INFO @ Fri, 02 Mar 2018 16:37:45: 6000000 INFO @ Fri, 02 Mar 2018 16:37:58: 7000000 INFO @ Fri, 02 Mar 2018 16:38:11: 8000000 INFO @ Fri, 02 Mar 2018 16:38:23: 9000000 INFO @ Fri, 02 Mar 2018 16:38:35: 10000000 INFO @ Fri, 02 Mar 2018 16:38:47: 11000000 INFO @ Fri, 02 Mar 2018 16:38:59: 12000000 INFO @ Fri, 02 Mar 2018 16:39:11: 13000000 INFO @ Fri, 02 Mar 2018 16:39:23: 14000000 INFO @ Fri, 02 Mar 2018 16:39:35: 15000000 INFO @ Fri, 02 Mar 2018 16:39:48: 16000000 INFO @ Fri, 02 Mar 2018 16:40:01: 17000000 INFO @ Fri, 02 Mar 2018 16:40:14: 18000000 INFO @ Fri, 02 Mar 2018 16:40:27: 19000000 INFO @ Fri, 02 Mar 2018 16:40:41: 20000000 INFO @ Fri, 02 Mar 2018 16:40:54: 21000000 INFO @ Fri, 02 Mar 2018 16:41:08: #1.2 read input tags... INFO @ Fri, 02 Mar 2018 16:41:21: 1000000 INFO @ Fri, 02 Mar 2018 16:41:34: 2000000 INFO @ Fri, 02 Mar 2018 16:41:47: 3000000 INFO @ Fri, 02 Mar 2018 16:42:00: 4000000 INFO @ Fri, 02 Mar 2018 16:42:13: 5000000 INFO @ Fri, 02 Mar 2018 16:42:26: 6000000 INFO @ Fri, 02 Mar 2018 16:42:39: 7000000 INFO @ Fri, 02 Mar 2018 16:42:52: 8000000 INFO @ Fri, 02 Mar 2018 16:43:05: 9000000 INFO @ Fri, 02 Mar 2018 16:43:17: 10000000 INFO @ Fri, 02 Mar 2018 16:43:31: 11000000 INFO @ Fri, 02 Mar 2018 16:43:44: 12000000 INFO @ Fri, 02 Mar 2018 16:43:57: 13000000 INFO @ Fri, 02 Mar 2018 16:44:10: 14000000 INFO @ Fri, 02 Mar 2018 16:44:23: 15000000 INFO @ Fri, 02 Mar 2018 16:44:35: 16000000 INFO @ Fri, 02 Mar 2018 16:44:49: 17000000 INFO @ Fri, 02 Mar 2018 16:45:02: 18000000 INFO @ Fri, 02 Mar 2018 16:45:14: 19000000 INFO @ Fri, 02 Mar 2018 16:45:40: #1 Background Redundant rate: 0.11 INFO @ Fri, 02 Mar 2018 16:45:40: #1 finished! INFO @ Fri, 02 Mar 2018 16:45:40: #2 Build Peak Model... INFO @ Fri, 02 Mar 2018 16:45:49: #2 number of paired peaks: 0 WARNING @ Fri, 02 Mar 2018 16:45:49: Too few paired peaks (0) so I can not build the model! Lower your MFOLD parameter may erase this error. WARNING @ Fri, 02 Mar 2018 16:45:49: Process is terminated!

this is the error report. I lower the MFOLD value to 10, and it still gives me the error. However, when I ask MACS to 'not build a shifting model', MACS is able to call the peaks, and I get apprx. 1650 peaks. Why is it like that? Does that mean that the peaks generated by MACS are not good peaks? Or could you explain it to me please? Thanks!

peaks parameters inputs macs macs2 • 337 views

ADD COMMENT • link •

modified 8 months ago • written 9 months ago by jenyka_dm • 10

9 months ago by

Jennifer Hillman Jackson ♦ 25k

United States

Jennifer Hillman Jackson ♦ 25k wrote:

Hello,

This is not a Galaxy wrapper tool problem (code), but rather a usage issue that is related to the inputs and parameters. In short, the data content processed with the given parameters is not able to call peaks.

Some troubleshooting places to start:

Use MACS2 instead of MACS (if not already). Be sure to use the latest version to avoid prior issues that have been resolved in the wrapper revisions.
Double check your alignment results. If a significant portion of your reads did not map, that can lead to errors like this one. The rerun and job details icons within any result dataset - even if in an error state - will report what was originally input.
- This includes a double check that the intended target genome was actually mapped against
- AND that the fastq data was properly prepared and has good enough quality to get mapped.
If mapping against a Custom genome, make sure it was formatted correctly. If not, fix it then remap, and rerun MACS2.
Make certain that these setting are a match for your input data and target genome build: effective genome size, tag size, band width.

FAQs covering most of the above checks and more help for troubleshooting are here: https://galaxyproject.org/support/#troubleshooting

Galaxy tutorials, including those for fastq QA/QC and ChIP-seq analysis: https://galaxyproject.org/learn

The MACS2 manual and google group are also good resources. Many Galaxy tools are wrappers around 3rd party tools and the scientific usage advice for command-line usage can be applied when using the same tool in Galaxy. See https://github.com/taoliu/MACS/blob/master/README.rst and the associated wiki help plus an example search for prior Q&A related to this warning about "too few paired peaks" at the tool's google group.

Thanks! Jen, Galaxy team

ADD COMMENT • link modified 9 months ago • written 9 months ago by Jennifer Hillman Jackson ♦ 25k

Thanks! If I keep using MACS but with the option of 'do not build shifting model' will that affect my downstream analysis in any way?

ADD REPLY • link written 8 months ago by jenyka_dm • 10

Yes, any differences in parameters between runs can result in differently called peaks. Please see the tool form help, including the link to the MACS2 documentation for how the options --nomodel, --shift, and --extsize impact analysis.

https://github.com/taoliu/MACS/blob/master/README.rst

ADD REPLY • link written 8 months ago by Jennifer Hillman Jackson ♦ 25k