STAR 2.5.2a hanging on gzipped input

305 views
Skip to first unread message

Margherita

unread,
Jun 8, 2016, 6:30:35 AM6/8/16
to rna-star
Dear Alex and RNA-STAR group,

First of all thank for your great work!!

And then the problem...
Yesterday I downloaded STAR version 2.5.2a (static Linux binary) to map some ChIP-seq data and I am having some problems with mapping gzipped input (the issue doesn't seem to be covered in previous threads and I personally have no clue).

Shortly, when specifying the option "--readFilesCommand gunzip -c" STAR hangs right after the command is issued and the same happens with "--readFilesCommand zcat". The mapping works fine if I extract independently the fastq and run on uncompressed input or if I use version 2.5.1b.

My command line is:
STAR --runThreadN 5 --genomeDir /media/margherita/DATAPART1/genomes/mm10/STAR_index/ --readFilesCommand zcat --readFilesIn test_R1.fa.gz test_R2.fa.gz --alignIntronMax 1 --alignEndsType EndToEnd --outFilterMultimapNmax 1 --outFileNamePrefix small_test


I am attaching a small input example and the Log of both failed attempts.

Best,
Margherita






test_data_and_logs.tar.gz

Alexander Dobin

unread,
Jun 10, 2016, 12:54:09 PM6/10/16
to rna-star
Hi Margherita,

I have tried running your fastq files with zcat and gunzip -c, and could not reproduce the problem.
If you run 2.5.1b exactly the same way (same server, paramteres, files, directories), does the problem occur?
This would be strange, as I have not made any changes to that part of the code from 2.5.1b to 2.5.2b .

Could you please do the following in the run directory:
$ mkfifo small_test_zcat__STARtmp/tmp.fifo.read1
$ zcat test_R1.fa.gz > small_test_zcat__STARtmp/tmp.fifo.read1 &
$ head small_test_zcat__STARtmp/tmp.fifo.read1
This should print first lines of the fastq file.

Cheers
Alex

Margherita

unread,
Jun 14, 2016, 11:25:58 AM6/14/16
to rna-star
Hi Alex,

Thank you very much for your prompt reply and your suggestions. You directed me towards the right direction, even though I still don't get the bottom-line problem (I suppose it's partition+mount-permissions related, not sure how).

In practice the fastq data I am trying to analyze is located in a 'data' partition which is mounted on /media/. When I tested 2.5.1b with the small example I shared, I had both the executable and the data in my home directory and I confirm that 2.5.2a works fine when the input data is in the home. I also see that 2.5.1b hangs when the input data is located in the mounted data partition. "zcat" by itself works fine wherever it's called from of course, as well as "gzip -c".

I imagine the changing the mount options is probably the solution of the problem, but is the behavior normal?

Thank you once more for your help and for STAR.
Margherita

Alexander Dobin

unread,
Jun 15, 2016, 12:53:39 PM6/15/16
to rna-star
Hi Margherita,

what happens if you try to do 
$ mkfifo /media/path/to/data/tmp1.fifo
Some partitions do not allow fifo files that are required for on the fly unzipping.

One way to circumvent this problem is to use --outTmpDir /home/path/to/temp/dir/ . Then the temp file (inlcuding fifo)  will be written into this directory.
Another option is to use process substitution
--readFliesIn <(zcat /path/to/read1.fq.gz) <(zcat /path/to/read2.fq.gz)
and omit --readFilesCommand option.

Cheers
Alex

Margherita

unread,
Jun 17, 2016, 11:40:34 AM6/17/16
to rna-star
Hi Alex,

If I issue, as you suggested, the command:
$ mkfifo /media/path/to/data/tmp1.fifo
the tmp1.fifo is created where specified.

Your command substitution suggestion works fine, so I will go with it.

Thank you very much for your help and sorry for bothering with something that doesn't seem to be STAR-related at all.

Cheers,
Margherita

Alexander Dobin

unread,
Jun 17, 2016, 12:06:40 PM6/17/16
to rna-star
Hi Margherita,

thanks for letting me know - not sure what the underlying problem is, but if the process substitution works, it's pretty much the same.

Cheers
Alex

Margherita

unread,
Jun 17, 2016, 12:27:01 PM6/17/16
to rna-star
Yup! Thanks :-)
Reply all
Reply to author
Forward
0 new messages