rietho / IPO

A Tool for automated Optimization of XCMS Parameters
http://bioconductor.org/packages/IPO/
Other
34 stars 20 forks source link

Issues running IPO on cluster with 15 files #66

Open adelabriere opened 4 years ago

adelabriere commented 4 years ago

Hi,

I am trying ot use IPO on 15 files on a cluster to speed up the process. However IPO keep crashing on our cluster. I tried 16 cores with 8 Go of RAM with should be more than enough for an optimization of 15 files. The BPPARAM Argument was a MulticoresParams with 15 cores.

After many DOE at some point we get the following message : Error in if (any(info$isdir)) { : missing value where TRUE/FALSE needed Calls: optimizeXcmsSet ... xcmsSetStatistic -> calculateXcmsSet ->

As this is happening after 24 successful iteration I guess it is coming from parallelism:

Here is the message in the last round, nothing special : 25 Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 9998 regions of interest ... OK: 789 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 11147 regions of interest ... OK: 687 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 11638 regions of interest ... OK: 1040 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 9688 regions of interest ... OK: 785 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 8930 regions of interest ... OK: 678 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 9507 regions of interest ... OK: 661 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 10268 regions of interest ... OK: 759 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 9005 regions of interest ... OK: 835 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 9606 regions of interest ... OK: 940 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 9554 regions of interest ... OK: 749 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 8143 regions of interest ... OK: 1344 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 10699 regions of interest ... OK: 1058 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 10607 regions of interest ... OK: 668 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 10095 regions of interest ... OK: 601 found. Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 10248 regions of interest ... OK: 999 found. 26 Detecting mass traces at 10.25 ppm ... OK Detecting chromatographic peaks in 9998 regions of interest ... OK: 794 found. Error in if (any(info$isdir)) { : missing value where TRUE/FALSE needed Calls: optimizeXcmsSet ... xcmsSetStatistic -> calculateXcmsSet -> In addition: There were 50 or more warnings (use warnings() to see the first 50) Execution halted

I hope that someone can help.

Best, Alexis

rietho commented 4 years ago

Hi @adelabriere! Unfortunately IPO us currently unsupported and I have very limited capacity to support.

However, I did look briefly into your error. I am assuming that the error is caused from a call to xcms::xcmsSet, namely from here. Here, file.info() is called on the files provided. I do assume, that one of your files provided is not pointing to an actual file. You might try to file.exists(files) on your files to check.

I hope this helps.

adelabriere commented 4 years ago

Hi @rietho , thanks for taking the time to answer.

Sadly this can t be the issue as my file are listed directly from a directory, I tried to process them using XCMS and it worked. All files exists and are readable. Moreover this error occurs at the 25th point of the design, and it successfully processed the 24 first, meaning that it is not a file issue.

I did not find a workaround at the moment. I m trying to rerun it with a single core on the cluster to see if it solves the issues.

Best, Alexis