Automatic 2-Pass segmentation fault

[craigc...@gmail.com]

Hi all,

I am attempting to use STAR 2.4j in automatic 2Pass mode (starting with unannotated genome and use the switches --twopass1readsN and --sjdbOverhang), but running get the following segmentation fault after I complete the first pass and it finished preparing junctions

and finished SA search (per the attached log files):

"home/ccb15/.lsbatch/1424722972.5239437: line 8: 15490 Segmentation fault"

I noticed in my log files (Attached) that there is an extra '/' in the SJ path:

Feb 23 15:38:57   Loaded database junctions from the 1st pass file: ./TAC_L1/TAC_L1._STARpass1//SJ.out.tab: 112762 junctions

Could that  be throwing the error?

Attached are my log files.  here is my command line (submitted to our cluster, with 12 cores and 72000 MB memory:

STAR --genomeDir STAR_1Pass_Genome_NoAnno --outFilterMultimapNmax 1 --outFilterIntronMotifs RemoveNoncanonical --runThreadN 12 --seedSearchStartLmax 25 --alignEndsType EndToEnd --outFilterMismatchNoverLmax 0.08 --outSAMtype BAM SortedByCoordinate --outSAMstrandField intronMotif --outStd BAM_SortedByCoordinate --readFilesCommand zcat --readFilesIn R1.fq.gz R2.fq.gz --twopass1readsN 40000000 --sjdbOverhang 49 --outFileNamePrefix ./TAC_L1/TAC_L1. > ./TAC_L1/TAC_L1.Accepted.out.bam

Any ideas? I also tried with the --limitBAMsortRAM switch, but same error. 

Thanks,

Craig

[Alexander Dobin]

Hi Craig,

I do not see anything suspicious in the Log.out file. Could you please send me the ./TAC_L1/TAC_L1._STARpass1//SJ.out.tab file, as well as a link to the genome fasta files.

Cheers

Alex

[craigc...@gmail.com]

I may have figured this out. I included the --sjdbOverhang switch at the genome generation stage as well as the mapping stage. On a different genome, I repeated the genome generation without the --sjdbOverhang switch, and mapping with the --sjdbOverhang switch, and it seems to be working now. Does that make sense to you?

Thanks

Craig