Bam SortedByCoordinate

[Gerben Menschaert]

Hi,

I'm trying the new functionality to output a 'sorted by coordinate' bam file.

While doing so, I ran into the following error:

##### Final effective command line:
STAR   --runThreadN 20   --genomeDir /Sequence/STARIndex/mmu.GRCm38.70.genome.35bpOverhang   --genomeLoad NoSharedMemory   --readFilesIn fastq/fastq1_norrna.fq      --outFileNamePrefix fastq1/   --outSAMtype BAM   SortedByCoordinate      --outSAMattributes All      --outFilterMultimapNmax 16   --outFilterMismatchNmax 2   --clip3pAdapterSeq CTGTAGGCACCATCAAT      --clip3pAdapterMMp 0.1      --seedSearchStartLmaxOverLread 0.5

Nov 07 14:11:21 ..... Started STAR run
Nov 07 14:15:36 ..... Started mapping
Nov 07 14:20:22 ..... Started sorting BAM
Max memory needed for sorting = 1097868417
EXITING because of FATAL ERROR: number of bytes expected from the BAM bin does not agree with the actual size on disk: 172355397   0   0
Nov 07 14:20:22 ...... FATAL ERROR, exiting

Many thanks in advance,

Gerben

[Alexander Dobin]

Hi Gerben,

apparently there are problems with scaling BAM sorting to a large number of threads.

I actually do not see any speed gain from using >6 threads (but it still faster than samtools with 12 threads).

I have pushed patch from GitHub master https://github.com/alexdobin/STAR where I added --outBAMsortingThreadN option to specify number of threads for BAM sorting. By default (0) it's equal to min(6,runThreadN).

If the error persists, please send me the Log.out file.

Cheers

Alex

[googl...@schwarzelan.de]

Hi Gerben,

I saw that, too, but only once.

After concluding my tests (see https://groups.google.com/forum/#!topic/rna-star/dr7Kb99E7OA ),

I successfully ran my STAR + Bam SortedByCoordinate with 60 threads without the error you saw - as did I in early runs.

I'm not sure if the error we see is a result from to-many-filehandles (which Alex did address with his latest changes) or something else,

we will have to see!

Best

-Jonathan

[Arang Rhie]

Hi Alex,

I get the similar message from that what Gerbin has reported:

Max memory needed for sorting = 10906470864

 

EXITING because of FATAL ERROR: number of bytes expected from the BAM bin does not agree with the actual size on disk: 3157610131   1974960457   1

Although I'm using runThreadN 6 and outBAMsortingThreadN 6.

I'm also using option chimSegmentMin 15, and the Chimeric.out.junction and Chimeric.out.sam got results in there.

However, Aligned.sortedByCoord.out.bam is empty.

This also happened when I'm using option BAM Unsorted.

I was expecting to get results from Chimeric.* and also the BAM file (since SAM file is too large in file size), together with SJ.out.tab,

but no SJ.out.tab is shown with the --chimSegmentMin option.

My server ulimit is set to 'unlimited', so I'm confused what's happening here.

Thank you in advanced,

Arang Rhie

2014년 11월 10일 월요일 오후 6시 31분 32초 UTC+9, googl...@schwarzelan.de 님의 말:

[Alexander Dobin]

Hi Arang,

this error might be cause by insufficient disk space, how much space do you have left before the run?

Please try the latest patch of STAR and send me the Log.out file if it still fails, as well as output of `ls -lR` on the STAR run directory.

Cheers

Alex