Closed jdmontenegro closed 3 years ago
Hi Juan,
Thank your for the detailed report of the issue.
Indeed, the problem seems to be related to the KmerInShort
executable needed by FEELnc for the fast extraction of k-mer profiles and which, a priori, does not work for your particular OS.
I'd suggest to try building KmerInShort
directly from the dedicated git repo and then add it to your PATH
as indicated in the Install section.
Best wishes,
Thomas
Thank you Thomas. The KmerInShort binary being used was compiled from the git repo, I didnt use any existing binaries.
Hi Juan,
Thanks also for the details issue. Did you try to run just the KmerInShort tool? To know if it is because of FEELnc+KmerInShort or just KmerInShort.
Best wishes, Valentin
Thank you @vwucher
I just tested KmerInShort alone and it seems to be working correctly.
$ KmerInShort --help
ERROR: Unknown parameter '--help'
ERROR: Option '-file' is mandatory
ERROR: Option '-kmer-size' is mandatory
[kis options]
-nb-cores (1 arg) : number of cores [default '0']
-verbose (1 arg) : verbosity level [default '1']
-version (0 arg) : version
-help (0 arg) : help
-file (1 arg) : input file
-kmer-size (1 arg) : ksize
-out (1 arg) : output file [default '']
-offset (1 arg) : starting offset [default '0']
-step (1 arg) : step [default '1']
-kval (1 arg) : file with kmer values [default '']
-dont-reverse (0 arg) : do not reverse kmers, count forward and reverse complement separately
-freq (0 arg) : output frequency
-perSeq (0 arg) : one output file and count per fasta sequence
-NSE (0 arg) : compute normalized Shannon entropy
-sum (0 arg) : compute sum over all files
so it does not seem to be an error here. Any specific command that I should try with KmerInShort?
Nope, wait I just ran a quick test and it segfaulted. Although not much information.
$ KmerInShort -file transcripts.busco.fa -kmer-size 21 > tmp
Counting in canonical mode (kmer and their reverse complement counted together)
Launching kmer counting with offset 0 step 1
[counting kmers ] 0 % elapsed: 0 min 0 sec remaining: 0 min 0 secSegmentation fault (core dumped)
I will recompile and test again.
thank you.
Ok. Thanks for letting us update! In case it is not working even after re-compiling, you can also try using the binary we did. Maybe it will work fine.
Bye, Valentin
The problem is still there, the test file for KmerInShort works properly, but my dataset keeps segfaulting. Seems like a bug in KmerInShort. Nobody has yet reported a similar error.
Ok. Did you try using the binary from the github folder? or only the one you compiled yourself? Also, one thing you can try is to split your data into 2, run KmerInShort on both, see for which one it is not working, then split this one into two, etc. To check if it is coming from a specific sequence.
Well my mistake, KmerInShort only uses kmer size up to 15 and I used 21, hence the segfault. When I run the same command with a k=15:
KmerInShort -file transcripts.fa -kmer-size 15 -nb-cores 8 > tmp
Counting in canonical mode (kmer and their reverse complement counted together)
Launching kmer counting with offset 0 step 1
[counting kmers ] 100 % elapsed: 0 min 2 sec remaining: 0 min 0 sec
There is no segfault. So KmerInShort is working correctly on its own. Any more suggestions would be very helpful.
Nope, wait I just ran a quick test and it segfaulted. Although not much information.
$ KmerInShort -file transcripts.busco.fa -kmer-size 21 > tmp Counting in canonical mode (kmer and their reverse complement counted together) Launching kmer counting with offset 0 step 1 [counting kmers ] 0 % elapsed: 0 min 0 sec remaining: 0 min 0 secSegmentation fault (core dumped)
I will recompile and test again.
thank you.
Ok.
So when you use KmerInShort alone, it is working fine. But then when you use FEELnc_codpot.pl it crashes, right? Did you try with your new compilation? and with the native binary?
Also, did you try to save the temporary file, using --keeptmp
and then check the files? The ones with which KermInShort is crashing? To see if it can come from this.
And trying the splitting of the input data can also help to find issues with the sequences if they exist.
Thank you for your suggestions, I switched to the KmerInShort version provided in the bin directory of FEELnc and the program ran correctly, so it seems to be something about the most recent version of KmerInShort or something broken during runtime in my local install.
Everything seems to be working properly now.
Best regards,
Juan D. Montenegro
Ok great.
I will close the issue then.
Regards, Valentin
Hello I just ran into this problem while trying to execute codpot:
I am getting a ton of segfaults, but I am not sure why. I am runinng on a linux Manjaro OS :+1: NAME="Manjaro Linux"
with perl5.32 and R4.0.3.
Any help would be more than welcome.
Kind regards,
Juan D. Montenegro