Alignment loop using STAR

Question

[lamia_203]

I'm running scRNA-seq and am trying to create a loop for all samples to run alignment. However, the loop stops about 20 or so samples later. Am I missing something from code? The alignment works fine without the loop.

for i in *.fastq.gz
do
    STAR --genomeDir ./human_index \
              --runMode alignReads \
              --readFilesIn $i \
              --runThreadN 10 \
              --readFilesCommand zcat \
              --outSAMtype BAM SortedByCoordinate \
              --chimSegmentMin 20 \
              --quantMode GeneCounts \
              --outReadsUnmapped Fastx \
              --outFileNamePrefix ./star_out/$i
done

EDIT by @RamRS: I formatted the code to be more readable. Original code copy-pasted below:

for i in *.fastq.gz; do STAR --genomeDir ./human_index --runMode alignReads --readFilesIn $i --runThreadN 10 --readFilesCommand zcat --outSAMtype BAM SortedByCoordinate --chimSegmentMin 20 --quantMode GeneCounts --outReadsUnmapped Fastx --outFileNamePrefix ./star_out/$i; done

[grant.hovhannisyan]

Look at the log files of the iteration at which the loop breaks. I suspect that you run out of memory when sorting the bam file.

[lieven.sterck]

there is a possibility that you're running out of storage space to store the results?

Did this not produce any errors then?

[lamia_203]

I had a slightly different code where the --runThreadN was set to 20 and it was aborting but now with it set to 10 its coming up with error message about not enough space for bam - however this time its continuing onto the next sample despite the error message

[h.mon]

Do you really need --outReadsUnmapped Fastx? Since you are running out of disk space (could you copy the exact error message?), you could save space by not outputting unmapped reads..

[lamia_203]

Jan 10 19:59:01 ..... started sorting BAM

EXITING because of FATAL ERROR: number of bytes expected from the BAM bin does not agree with the actual size on disk: 4944559   4928845   15

    Jan 10 19:59:03 ...... FATAL ERROR, exiting

    EXITING because of FATAL ERROR: number of bytes expected from the BAM bin does not agree with the actual size on disk: 4653064   4637278   13

    Jan 10 19:59:03 ...... FATAL ERROR, exiting
    *** Error in `STAR': double free or corruption (!prev): 0x00000000026d89a0 ***
    ======= Backtrace: =========
    /lib/x86_64-linux-gnu/libc.so.6(+0x777e5)[0x7f08711c27e5]
    /lib/x86_64-linux-gnu/libc.so.6(+0x8037a)[0x7f08711cb37a]

    7ffdd9196000-7ffdd9199000 r--p 00000000 00:00 0                          [vvar]
    7ffdd9199000-7ffdd919b000 r-xp 00000000 00:00 0                          [vdso]
    ffffffffff600000-ffffffffff601000 r-xp 00000000 00:00 0                  [vsyscall]
    *** Error in `STAR': corrupted double-linked list: 0x00000000026d8990 ***
    Aborted (core dumped)
   Jan 10 19:59:07 ..... started STAR run

The error comes after sorting out the bam files. I previously used --outReadsUnmapped Fastx? and thought to keep it incase I wanted to know the reads that were unmapped. That is a good idea to save space, thank you.

Answer 1

[swbarnes2]

It would help to know what the error is. If the issue isn't storage space, it might be RAM used while sorting. You might try adding this

--limitBAMsortRAM 31000000000

To your STAR command line.

[lamia_203]

I added this code and it worked overnight, thank you! Just to confirm, is this limiting the size of the BAM files? or is it doing something else?

[GenoMax]

It is limiting amount of RAM used for sorting to 31GB.