RNAstar alignment segmentation fault

Question

[sam]

Hello,

I have a question regarding the RNA-star alignment program. I'm trying to map thousands of reads (~7000 sequences) against a fasta file containing a handful of sequences (~20 sequences). In order to do that in RNA-star, I had to generate a genome index based on the small fasta file that contains ~20 sequences:

$STAR --runMode genomeGenerate --genomeDir $direc --genomeFastaFiles $filename --runThreadN 4

Next I run the following command to do the mapping:

$STAR --genomeLoad NoSharedMemory --outFileNamePrefix $direc/STAR. --genomeDir $direc --readFilesIn file.fasta

After running the above I get the following segmentation fault error:

Jul 22 11:44:49 ..... Started STAR run
Jul 22 11:44:50 ..... Started mapping
Segmentation fault (core dumped)

I think this has something to do with the size of the genome index I'm generating. This is because I was able to run the above commands successfully when I had a larger number of sequences used to generate the genome index. Has anyone run into this issue before? If yes, any ideas/suggestions on how to resolve it?

Answer 1

[sam]

I found what was the problem. When generating a genome index, this parameter must be changed to accommodate smaller genomes:

genomeSAindexNbases 14

int: length (bases) of the SA pre-indexing string. Typically between 10 and 15. Longer strings will use much more memory, but allow faster searches.

[Ron]

Hi Sam,

I have tried implementing the parameter genomeSAindexNbases 14,but gives me the same segmentation error (core dumped).I have increased memory to 40G but sometimes it runs and sometimes it does not give the output results.It seems to be random.

Do you think there could be any other reason?

Best,
Ron

[Ding Wenchao]

Hi Ron,

genomeSAindexNbases 14 is the default setting, you should change this parameter according to the genome size.

min(14, log2(GenomeLength)/2 - 1)

[amra.dhabalia]

You have to use the following: genomeSAindexNbases <integer> while making the index of the small genome