GstrandBit problem

[Zeran Li]

Hi Alex,

I am using Star 2.5.2b version to generate second index for 2-pass. I keep having this GstrandBit error.
My reads are 2x151 PE and I have also tried to lower --sjdbOverhang from 150 to 100 but still have this error.

Here is the command I run:
STAR-2.5.2b/bin/Linux_x86_64_static/STAR --runThreadN 22 \
--runMode genomeGenerate \
--genomeDir Star_index_lite_spike-in_pass2_2.5.2b \
--genomeFastaFiles GRCh37.p13.genome.lite.spike-in.fa \
--sjdbGTFfile  gencode.v19.annotation.spike-in.tab.gtf \
--sjdbOverhang 100 \
--sjdbFileChrStartEnd Pass1/Pass1_SJ.out.tab \
--limitSjdbInsertNsj 6000000 \
--limitGenomeGenerateRAM 90953375957 \
--sjdbInsertSave All

I concatenate all SJ.out.tab files from pass1 and created this Pass1_SJ.out.tab file. I also have spike-in sequence in my fasta and gtf.
What's interesting is that previously I used full version of Gencode v19 which have all the alternative haplotypes, and I was able to finish 2-pass without any problem. Then I realized I shouldn't use the full version but the lite version. Because gencode v19 don't have primary assembly I have to create my own by removing all the alternative loci, and then for the second pass genomeGenerate step I start to have this GstrandBit error.

Attached is my Log.out for your reference.

Best regards,
Zeran

[Alexander Dobin]

Hi Zeran,

this problem is caused by the genome+junctions length being close to the 2^32 boundary.

To avoid this problem you would need to specify more precisely the total number of (collapsed) junctions to be inserted in the genome.

You can find this number in the Log.out file:

Jul 28 10:04:01   Finished SA search: number of new junctions=3292413, old junctions=0

If you specify --limitSjdbInsertNsj 3400000, the error should go away.

Cheers

Alex

[Zeran Li]

Alex,

Problem solved with your suggested changes.

Thank you so much!

Best regards,
Zeran