Calcbmr Of Music(0.4) Error: Error Runing Command Joinx1.6

Question

[Ryo Murasaki]

Hi all,

I am interested in using the Genome MuSiC tool from WashU to calculate Mutation P-value.

I got the following errors in CalcBmr of MuSiC 0.4(current release) after CalcCovg running.

input command is below:

genome music bmr calc-bmr \
--bam-list ${BAM_LIST} \
--maf-file ${MAF_FILE} \
--output-dir ${OUTPUT_DIR} \
--reference-sequence ${REF_SEQ} \
--roi-file ${ROI_FILE} \
1>${LOG_CALC_BMR} \
2>&1;

Log & error messege is below:

Loading per-sample coverages stored in /home/ryo/output/total_covgs
Loading per-gene coverage files stored under /home/ryo/output/gene_covgs/
Running 'joinx1.6 ref-stats' to read reference FASTA and identify SNVs at AT, CG, CpG sites
ERROR: ERROR RUNNING COMMAND. Failed to execute: joinx1.6 ref-stats --ref-bases --bed /tmp/gm-genome_sys-2012-10-18_19_56_46--hFMV/anonymous0 --fasta /home/ryo/fasta/d5cs.fasta --output /tmp/gm-genome_sys-2012-10-18_19_56_46--hFMV/anonymous1 at /usr/share/perl5/Genome/Model/Tools/Music/Bmr/CalcBmr.pm line 324

Could somebody explain the above problems? Thank you.

[Chris Miller]

Do you have joinx installed? It's a music dependency. Is it in your path? (what happens if you type "joinx" from a command line?)

Answer 1

[Ryo Murasaki]

Hi all,

I solved it.

Ubuntu's Memory( or swap) is too low to exec Genome::Sys->shellcmd ( exec system() in Perl) in CalcBmr.pm.

For example, This linesCalcBmr.pm) need high memory volume.

279     foreach my $gene ( @all_gene_names ) {
280         foreach my $sample ( @all_sample_names ) {
281             $gene_mr[$sample_idx{$sample}][$gene_idx{$gene}][$_][mutations] = 0 foreach( @mut_classes );
282             $gene_mr[$sample_idx{$sample}][$gene_idx{$gene}][$_][covd_bases] = 0 foreach( @mut_classes );
283         }
284     }

I turned up swap volume.

sudo dd if=/dev/zero of=/swap.extended bs=1G count=10
sudo mkswap /swap.extended
sudo swapon /swap.extended

so, I can exec CalcBmr!

Thank you all,

Ryo

Answer 2

[Cyriac Kandoth]

In general, if you have problems with MuSiC's bmr calc-bmr...

1. Ensure that the MAF file, reference fasta, BAM files, and ROI file, all use the same chromosome naming conventions and the same genomic bounds. For example, if your BAMs and reference fasta use the chr-prefix for chromosome names, then the MAF and ROI files should use the same. And definitely make sure you don't mix hg18 with hg19. MAF column 4 specifies which reference build is in use.
2. Memory requirements are quite expensive for bmr calc-bmr, but it increases fairly linearly with more samples in your cohort. For example, I needed 3.9 GB RAM for a cohort of 98 cases, and 113.3GB for 3210 cases.
3. bmr calc-bmr uses the gene names in column 1 of the MAF to match variants to their respective ROIs. So ensure that the annotation database used to construct your ROI file is the same database used to annotate your MAF file.

Answer 3

[Ryo Murasaki]

Hi Chris

thank you very much for your advice.

I have installed joinx. and I tried below command to check.

murasaki@ub01:tmp$ which joinx1.6
/usr/bin/joinx1.6

and

murasaki@ub01:tmp$ dpkg -L joinx1.6
/usr
/usr/share
/usr/share/joinx1.6
/usr/share/joinx1.6/VcfReport.R
/usr/share/man
/usr/share/man/man1
/usr/share/man/man1/joinx1.6.1.gz
/usr/share/doc
/usr/share/doc/joinx1.6
/usr/share/doc/joinx1.6/copyright
/usr/bin
/usr/bin/joinx1.6

I don't would like to remove tmp directory and tmp file(anonymous file), so I rewrited 'my $cleanup=0' at /usr/share/perl5/Genome/Sys.pm 145

After above error, I executed below(command in error msg).

joinx1.6 ref-stats --ref-bases --bed /tmp/gm-genome_sys-2012-10-18_19_56_46--hFMV/anonymous0 --fasta /home/ryo/fasta/d5cs.fasta --output /tmp/gm-genome_sys-2012-10-18_19_56_46--hFMV/anonymous1

so, this command ran successfully in bash, and Annonymous1 file is made successfully.

but calcbmr don't run yet..

[Alastair Droop]

Hello.

Although this post is 6 months old and no answers yet, I'm hoping someone can solve it. I'm having exactly the same issue.

joinx exists, and I'm running calc_covg just fine. calc_bmr dies:

Loading per-sample coverages stored in /home/medapd/Desktop/music-0634LEE/music/calc-covg/total_covgs
Loading per-gene coverage files stored under /home/medapd/Desktop/music-0634LEE/music/calc-covg/gene_covgs/
Running 'joinx1.6 ref-stats' to read reference FASTA and identify SNVs at AT, CG, CpG sites
bash: line 1: 12382 Segmentation fault      (core dumped) joinx1.6 ref-stats --ref-bases --bed /tmp/gm-genome_sys-2013-04-29_10_41_26--7hVz/anonymous0 --fasta /home/medapd/Desktop/music-0634LEE/Reference/GATK_g1k/human_g1k_v37.fasta.gz --output /tmp/gm-genome_sys-2013-04-29_10_41_26--7hVz/anonymous1
ERROR: ERROR RUNNING COMMAND.  Exit code 139, msg No such file or directory from: joinx1.6 ref-stats --ref-bases --bed /tmp/gm-genome_sys-2013-04-29_10_41_26--7hVz/anonymous0 --fasta /home/medapd/Desktop/music-0634LEE/Reference/GATK_g1k/human_g1k_v37.fasta.gz --output /tmp/gm-genome_sys-2013-04-29_10_41_26--7hVz/anonymous1 at /usr/share/perl5/Genome/Model/Tools/Music/Bmr/CalcBmr.pm line 324

I've tried setting the environment variable: export GENOME_SYS_NO_CLEANUP, but this just makes calc_bmr complain and doesn't help.

Extra swap doesn't help either.

Any ideas?

Thanks, Alastair

[Cyriac Kandoth]

Alastair, looks like you're using a reference fasta in .gz format. Try using it unpacked.

[Alastair Droop]

That's sorted it!

Thanks very much! Alastair