Segmentation fault short_read_analysis

[pguenzi-tiberi]

Hello everyone

I'm trying to use OPERA-MS to assemble a metagenome. Here is my command line:

=====
/bettik/guenzitp/mambaforge_installation_new/bin/perl /bettik/guenzitp/results/metag_operams/OPERA-MS/OPERA-MS.pl --short-read1 /bettik/guenzitp/data/XXXX/RQ/MiSeq/XXXX_RQ_tr_1.fastq --short-read2 /bettik/guenzitp/data/XXXX/RQ/MiSeq/XXXX_RQ_tr_2.fastq --long-read /bettik/guenzitp/data/XXXX/CR_2024_no_filter/Minion/nanopore.fastq --no-ref-clustering --out-dir results_assembly_metag --num-processors 8

=====

It stops after step 3 with this error:

Error in during cov_estimate.pl. Please see /bettik/guenzitp/results/metag_operams/results_assembly_metag//intermediate_files/coverage_estimation/cov_estimate.out and /bettik/guenzitp/results/metag_operams/results_assembly_metag//intermediate_files/coverage_estimation/cov_estimate.err

it's a problem with the short read pre-processing:

Error during short read pre-preocessing. Please see /bettik/guenzitp/results/metag_operams/results_assembly_metag//intermediate_files/coverage_estimation/preprocess_reads.err /bettik/guenzitp/results/metag_operams/results_assembly_metag//intermediate_files/coverage_estimation/short_read_analysis.out /bettik/guenzitp/results/metag_operams/results_assembly_metag//intermediate_files/coverage_estimation/short_read_analysis.err for details.

This is what is written in the “preprocess_reads.err” file:

Segmentation fault
[M::bwa_idx_load_from_disk] read 0 ALT contigs
[M::process] read 695800 sequences (200000006 bp)...
[M::process] read 694212 sequences (200000568 bp)...
[M::mem_process_seqs] Processed 695800 reads in 620.534 CPU sec, 79.632 real sec
[samopen] SAM header is present: 5119717 sequences.

I think I have a problem with the “short_read_analysis” program. When I run “./short_read_analysis”, I get a "segmentation fault" error.

What should I do?

Thanks in advance!

Pierre Guenzi-Tiberi

[jsgounot]

Hi Pierre,

do you have anything written in the short_read_analysis.out or short_read_analysis.err files? There is also a chance that you get this error because your system ran out of memory, could you try to rerun the script while checking the memory consumption?

Regards,
Jean-Sebastien

[pguenzi-tiberi]

Both files (short_read_analysis.out and short_read_analysis.err) are empty...
I immediately thought it might be a memory problem, so I tried with much more memory (about 500GB) and got the same error. I then tried to run the program without arguments, just to see what happened and, once again, I got the same error.
What's more, I get the same error with the test files, so I think the problem does indeed come from the short_read_analysis program.

Thanks for your help!
Pierre

[jsgounot]

Oh I see, thanks for checking all this out. Do you have anything in your preprocess_reads.out file? This could help me understanding what's happening here. Could you also check that the bwa and samtools in your tools_opera_ms folder are not raising a segfault when you print a help message? Thanks.

[pguenzi-tiberi]

There is no “preprocess_reads.out” file. Only a “preprocess_reads.err” file is created, where it is written :

Segmentation fault
[M::bwa_idx_load_from_disk] read 0 ALT contigs
[M::process] read 695800 sequences (200000006 bp)...
[M::process] read 694212 sequences (200000568 bp)...
[M::mem_process_seqs] Processed 695800 reads in 637.620 CPU sec, 81.630 real sec
[samopen] SAM header is present: 5119717 sequences.*

Also, samtools and bwa stored in the “tools_opera_ms” directory seem to work fine. I tried indexing a genome with bwa and using samtools view to convert a file and it worked.

[pguenzi-tiberi]

I have tried running OPERA-LG, OPERA-LG -fast and sigma (the other compiled programs) and nothing works. Every time I get the “segmentation fault” error. But when I tried to compile OPERA-LG from its own repository (https://sourceforge.net/projects/operasf/files/OPERA-LG%20version%202.0/), OPERA-LG works...

[jsgounot]

Do you have a compilation error when compiling OPERA-MS tools (OPERA-LG and Sigma). You can find the compilation commands here. If you don't, can you retest the two softs after you compiled those again?

[pguenzi-tiberi]

After the git clone command, I tried to compile everything. I just got an error at the end because OPERA-LG doesn't work even though I compiled it without error.
You can see the compilation results logout_compile_operams.txt

[jsgounot]

Ok, this one will be hard to fix for me, as it seems the C++ code is not working properly on your configuration. Would you mind sharing your system settings (OS, distribution and G++ or Clang++ versions)? You could also try to create a new environment with a different gcc version (like gcc 4.8.3) and compile again. I won't be able to reproduce the bug w/o having access to the same environment, so I'd suggest trying OPERA-MS on a different cluster / computer if you can.