SRA to fastq

[tinguzman]

Hi,

I'm trying to convert an sra file to fastq. i downloaded the sra toolkit and ran fastq-dump, here's what i got:


"Unrecognized character \xCF; marked by <-- HERE after <-- HERE near column 1 at fastq-dump line 1."

can somebody help me, are there any other script to convert sra files to fastq?

thanks,
christine

[WhatsOEver]

Hi christine,
can you provide the command you used and where you got the SRA file from. I don't think that fastq-dump is the problem, but rather your SRA file.

[tinguzman]

Hi

thanks for your reply. I ran this command, within the bin directory, - perl fastq-dump.2.3.5.2 SRR504687.sra. I also copied the sra file to the bin.

I downloaded this rsa file from here http://www.ncbi.nlm.nih.gov/sra/SRX151862%5Baccn%5D

thanks,
Christine

[WhatsOEver]

perl fastq-dump.2.3.5.2 SRR504687.sra.

That's the problem. What you have is a pre-compiled binary, not a perl script. Simply running

./fastq-dump.2.3.5.2 ./SRR504687.sra

from within your bin directory should work.

[tinguzman]

Hi,

thanks a lot! it's working now.

best,
christine

[tinguzman]

hi again!

i just finished running ./fastq-dump.2.3.5.2 -split-files ./SRR504687.sra. i have 2 output SRR504687_1 and SRR504687_2. one is 13gb while the other is only 3gb. i'm expecting that they should have the same size, for forward and reverse reads, right? correct me if i'm wrong coz i'm planning to assemble them using trinity. should I qc them first?

best,
christine

[dpryan]

You can download the compressed fastq files from ENA (here and here). That's often faster than dealing with SRA, since the SRA toolkit is painfully slow and often buggy. You'll note that their fastq files also have a different size, so this isn't something you're doing wrong.

[tinguzman]

Hi Ryan,

thank you very much. submitted fastq files or sra files are not yet trimmed/filtered right? they are raw sequences, for example from illumina sequencing.

best,
christine

[dpryan]

Yeah, they should be the raw files. This definitely raises the question of why the files have such different sizes. My guess is that the only person that can really answer that is the person that uploaded the files (Ana Riesgo).

[WhatsOEver]

hi again!

i just finished running ./fastq-dump.2.3.5.2 -split-files ./SRR504687.sra. i have 2 output SRR504687_1 and SRR504687_2. one is 13gb while the other is only 3gb. i'm expecting that they should have the same size, for forward and reverse reads, right? correct me if i'm wrong coz i'm planning to assemble them using trinity. should I qc them first?

best,
christine

I should have mentioned that the -split-3 command is always preferred over "split-files". This way 1 file will be generated for single end data, 2 files (with sufffix "_1" and "_2") for paired ends and 3 files (suffix "_1", "_2" and w/o suffix) if there are reads without mate pairs. This way I get a 13.2Gb file ("_1") and a 7.0Gb file ("_2") from the sra.
Looking at the fastq files, you'll see that forward reads are 150bp while reverse are 48bp. I have actually never seen this for PE data before and would contact the uploader (as dpryan suggested). At least there are no missing mates... I would give it a try and use them in trinity while waiting for an answer.

[lac302]

Has anyone had any issues downloading and splitting files from dnanexus? I'm unable to use aspera connect to download files from sra. I'm trying to replicate an experiment to test out an RNA-seq analysis pipeline but am having issues with 3 particular files. SRR639124, SRR639239, SRR639263.

The sra toolkit is working as other files has split without issue. SRR639124 oddly splits to two files _1 is 100bp and _2 is only 1bp? The other two will not split. The stat script shows only 1 read.

I'm wondering if it's an issue with dnanexus or the original SRA upload. Any help would be appreciated.

[GenoMax]

Are you using the latest sratoolkit? You no longer need to download .sra files. fastq-dump from the new sratoolkit will download fastq files directly.

[lac302]

Are you using the latest sratoolkit? You no longer need to download .sra files. fastq-dump from the new sratoolkit will download fastq files directly.

I believe so. Using 2.4.3.

How exactly does that work? curl?

[GenoMax]

I believe so. Using 2.4.3.

How exactly does that work? curl?

See this: http://seqanswers.com/forums/showpos...36&postcount=7

BTW: SRR639124 appears to be a single end dataset irrespective of what the SRA record says (this type of thing happens at times).

[lac302]

Thanks Geno. I need to improve my search skills.