Fastq-dump crashing with core dump?

Question

[alan]

I've got a bunch of SRR files that I need to convert into fastqs. I'm running a command like this one:

~/bin/sratoolkit.2.5.1-centos_linux64/bin/fastq-dump \
  -v -v \
  --split-3 \
  --outdir /<OUTPATH>/ \
  /<SOMEPATH>/SRR988439.sra.ncbi_enc

And a bunch of the time, it's crashing with a core dump and generating an ncbi_error_report.xml.

Here's the only segment there that seems to show an error, but I can't figure what's going on.

 <SOFTWARE>
  <VDBLibrary vers="2.7.2"/>
  <Build static="false">
   <Error rc="RC(rcCont,rcNamelist,rcInserting,rcString,rcEmpty)" function="VDBManagerListExternalSchemaModules"/>
  </Build>
  <Tool date="May 28 2015" name="/home/<SOMEUSER>/ncbi/sratoolkit.2.5.1-centos_linux64/bin/fastq-dump" vers="2.5.1">
   <Binary path="/home/<SOMEUSER>/ncbi/sratoolkit.2.5.1-centos_linux64/bin/fastq-dump" type="alias" md5="1a8642e8d30d6c6cfd75cccc383cc70b">
    <Alias resolved="fastq-dump.2">
     <Alias resolved="fastq-dump.2.5.1"/>
    </Alias>
   </Binary>
  </Tool>
 </SOFTWARE>

I would also note that I have had this same error with an earlier version of the SRA toolkit. (2.3.5)

[Ram]

What do you get when you type this in your home folder:

which fastq-dump

[alan]

fastq-dump isn't on my path.

[Ram]

Just checking.

Answer 1

[GenoMax]

That is an encrypted SRA file. You will need to decrypt it using http://www.ncbi.nlm.nih.gov/Traces/sra/sra.cgi?view=toolkit_doc&f=vdb-decrypt

This procedure will probably work: http://seqanswers.com/forums/showpost.php?p=102478&postcount=2

[alan]

But why would it take hours to fail? Why would it generate multi-gigabyte partial (?) fastqs that look fine except for possible truncation?

http://www.ncbi.nlm.nih.gov/books/NBK63512/#Download.are_downloaded_files_encrypted

The SRA (short-read-archive) data distributed through the dbGaP are encrypted but there is no need to decrypt them. The NCBI SRA toolkit can work directly on encrypted SRA data without decryption. Decrypted SRA data is in a binary format that is not human readable and can only be processed by the SRA toolkit anyway.

[GenoMax]

That guide is a bit vague. You do have the dbGaP repository key for your account since that seems to be required? Perhaps someone who has direct working experience with these type of files will be able to help.