Read correction jobs failed #1732
Comments
|
Not sure, never seen this error before. Are you able to share the data (either the raw input or the seqStore and corStores). |
|
Thank you for your response, I shared my raw data at ftp. |
|
Actually not all partition failed, 34 from 64 batches was successful. -- Opening seqStore '../../canu_min_q5.seqStore'. |
|
I took a look at your fastq file, it seems the upload is truncated or the gzip is corrupt, there is no quality or new line for the last read: It seems this is only about 1/2 of your data. It also looks like the vast majority of the data is extremely short, <500bp, of the 2m reads in the uploaded file, 1.6m are ignored for length and pretty much all reads are <3kb. I don't think you're going to get much assembly given such a poor read length distribution. At least with this set correction ran with no problems on both the 2.0 release and tip. Can you try uploading it again? |
|
I am sorry, I already re-uploaded it (total fastq data 7.1 Gb). |
|
I think this is an issue with the POSIX build. I tried both the tip and the 2.0 version and it ran without error to the end. The POSIX option generated a much larger corStore and was running slower as well. If you're able to build/run w/o the POSIX option that will likely fix this issue while we look into it. I can also share the corrected reads/assembly I generated locally. |
|
Thank you for your advice. |
… object store methods from meryl-utility. Issue #1732.
|
This has been resolved. The POSIX compliant names were causing some statistics on overlaps to be lost (due to a filename collision), which then resulted in far, far, far too many overlaps being used to correct reads - hence the ginormous corStore and out of memory errors. Sadly, since the overlap store is missing data, you'll need to start a new assembly. Note that the POSIX_FILE_NAMES option is no longer needed. |
Hi developers.
I am using canu from latest source code, with make POSIX_FILE_NAMES=1.
and ran command ~/canu -d canu_min_q5_2/ -p canu_min_q5 genomeSize=10m -nanopore q5.fastq corMinCoverage=0 corOutCoverage=all corMhapSensitivity=high correctedErrorRate=0.16 redMemory=60 oeaMemory=60 batMemory=60 gridOptionsovs=-l obtovlMemory=60 ovsMemory=60 corMemory=60 gridOptions=--time=240:00:00
and I got failed in read correction after 3 days run.
My read is metagenome
This is inside correctReads.000001.out
Found perl:
/home/wakimoto/Programs/anaconda3/bin/perl
This is perl 5, version 26, subversion 2 (v5.26.2) built for x86_64-linux-thread-multi
Found java:
/home/wakimoto/Programs/anaconda3/bin/java
openjdk version "1.8.0_152-release"
Found canu:
/home/wakimoto/Programs/canu/Linux-amd64/bin/canu
canu snapshot v2.0-development +532 changes (r10025 14fca27)
Running job 1 based on command line options.
And this is inside 0001.err
Any Advice?
The text was updated successfully, but these errors were encountered: