Can STAR work with compressed .gz FASTQ input?

Question

[biomagician]

Dear all,

I would like to align some reads to a reference using STAR. The following command works perfectly:

STAR --genomeDir output/spiking/index/star --readFilesIn reads.fastq --outFileNamePrefix outputFolder --runThreadN 8 > message.txt

However, the following command does not work:

STAR --genomeDir output/spiking/index/star --readFilesIn reads.fastq.gz --outFileNamePrefix outputFolder --runThreadN 8 > message.txt

Error: the read ID should start with @ or >

Presumably, STAR expects a FASTQ file and not a fastq.gz file. Does anybody know how to get round this in an efficient way?

Thanks.

C.

Answer 1

[Santosh Anand]

add --readFilesCommand zcat in your commandline

[biomagician]

Thanks. That didn't work for me probably because I am on Mac OSX. But I've found this and 'gunzip -c' instead of 'zcat' worked. https://groups.google.com/forum/#!topic/rna-star/NY56kU3mC64

[Santosh Anand]

Actually, you can use the bash shell hack <(gunzip -c filename.gz) to pass the gzipped file (or similarly, any other kind of zip file), which doesn't have a built-in mechanism to read the zipped files directly (STAR is awesome in providing the built-in mechanism :). It uses a trick of shell called Process Substitution. In essence, the command inside () is run inside a subshell, and the output of the command is passed as a input to the main command (which is your main program).

[biomagician]

Hi Santosh, Thanks for the reply. I am afraid this sounds a bit too technical for me. Do you know how the whole command would look like using this bash shell hack? C.

[Santosh Anand]

Suppose STAR was not supporting the gzipped file (reads.fastq.gz), then you would have run STAR like this: STAR --genomeDir output/spiking/index/star --readFilesIn <(gunzip -c reads.fastq.gz) --outFileNamePrefix outputFolder --runThreadN 8 > message.txt . Essentially, you need to put the zipped-file with the unzipping commands inside <()

The same mechanism for any other program which doesn't support zipped file natively.

[biomagician]

Thanks, I didn't know about this!