core dump on mapping

[GoogleCodeExporter]

STAR 2.3.0e
Linux dna 3.2.0-41-generic #66-Ubuntu SM

THIS WORKED:
* The fasta file is a 2.8 Mbp bacteria

 STAR --runMode genomeGenerate --genomeDir ref --genomeFastaFiles 6008.fna --runThreadN 32

THIS CORE DUMPED:
* Read file is FASTQ, 31bp reads, in Phred+64 format.

Core was generated by `STAR --genomeDir ref --readFilesIn 6008_mRNA.fastq 
--runThreadN 32'.
Program terminated with signal 11, Segmentation fault.
#0  0x0000000000405e4b in compareSeqToGenome(char**, unsigned long long, 
unsigned long long, unsigned long long, char*, PackedArray&, unsigned long 
long, bool, bool&, Parameters*) ()

Original issue reported on code.google.com by torsten....@gmail.com on 6 Jun 2013 at 6:07

[GoogleCodeExporter]

I went to the forums and dug around. 

It seems to work if I add "--genomeSAindexNbases 2" to the index generation 
step, as suggested by another user for small genomes.

I still consider a core dump a bug however :-)

Could you stat() the genome files to estimate the genome sizes and auto-choose 
good parameters to avoid this problem?

Original comment by torsten....@gmail.com on 6 Jun 2013 at 6:26

[GoogleCodeExporter]

Hi Torsten,

this is indeed a bug and hope to have a fix ready for the next release in ~July.

Thanks, Alex

Original comment by adobin@gmail.com on 8 Jun 2013 at 2:31

• Changed state: Started