search

mikewolfe / ChIPseq_pipeline

A general ChIP-seq pipeline to reproducibly process many samples at once.
0 stars 3 forks source link

Error in macs2_peakcalling on test #16

Closed cmhustmyer closed 3 years ago

cmhustmyer commented 3 years ago

Hi Mike--

We are running the test data of the ChIPseq_pipeline and we keep getting an error in the macs2 peak calling. When I check the error log this is what we find:

/results/peak_calling/logs/macs2/ Traceback (most recent call last): File "/mnt/scratch/colton/ChIPseq_pipeline/.snakemake/conda/1acff218/bin/macs2", line 653, in main() File "/mnt/scratch/colton/ChIPseq_pipeline/.snakemake/conda/1acff218/bin/macs2", line 49, in main from MACS2.callpeak_cmd import run File "/mnt/scratch/colton/ChIPseq_pipeline/.snakemake/conda/1acff218/lib/python3.8/site-packages/MACS2/callpeak_cmd.py", line 23, in from MACS2.OptValidator import opt_validate File "/mnt/scratch/colton/ChIPseq_pipeline/.snakemake/conda/1acff218/lib/python3.8/site-packages/MACS2/OptValidator.py", line 20, in from MACS2.IO.Parser import BEDParser, ELANDResultParser, ELANDMultiParser, \ File "init.pxd", line 242, in init MACS2.IO.Parser ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject

I do not get this error on my PC, but Colton is on a mac, not sure if that's why? We have bioconda installed, I also updated Bowtie and macs_2 just in case but that doesn't seem to help. Attached is a pic of the error in the pipeline

macs2_error

The config and pep files are all the test files: we did not make any changes. Thanks.

mikewolfe commented 3 years ago

Hey Christine, that is strange. Can you confirm for me that you are running this on the linux server and not on Colton's local computer?

cmhustmyer commented 3 years ago

Hi Mike,

Yes, he is running the pipeline on the Landick-1 server in his scratch folder, not on his local computer. You can see in the attached photo, too, that he’s logged into scratch and his biochem account in the margin to the left.

I thought maybe it that he needed some sort of additional program to run it (even though that doesn’t make much sense), so I also had him install your pytools program from github. We later tried to install bioconda, bowtie2, macs, etc. Nothing helped and he’s gotten this same error every time he tries to run the pipeline.

He can run the independent modules (preprocessing, alignment, coverage and norm, and WC) without any issue, it’s just peak_calling.

I can run peak_calling without any issue, I don’t know why this issue is occurring with the test module.

Thanks, CH

From: Mike Wolfe @.> Sent: Thursday, April 1, 2021 12:39 PM To: mikewolfe/ChIPseq_pipeline @.> Cc: Christine Hustmyer @.>; Author @.> Subject: Re: [mikewolfe/ChIPseq_pipeline] Error in macs2_peakcalling on test (#16)

Hey Christine, that is strange. Can you confirm for me that you are running this on the linux server and not on Colton's local computer?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/mikewolfe/ChIPseq_pipeline/issues/16#issuecomment-812064076, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ASRQ24ZTKWPNQYKEIE3FKL3TGSVTRANCNFSM42HHNHNQ.

mikewolfe commented 3 years ago

I was able to replicate the issue on a fresh install of the pipeline. Upgrading numpy to version 1.20 in the peak calling environment fixes the issue. It appears that there is some weird incompatibility between the unspecified library defaults and the older version of numpy.

Submitting a pull request to fix it shortly.