Splitting 454 paired reads in a FASTQ file

[maubp]

It certainly would be nice if the SRA allowed you download SFF files, seeing as that is how they are uploaded.

I've heard that they used to, and I agree it would make life easier it they still did - particularly for paired end 454 data.

[kmcarr]

[SLAP_HEAD]Doh![/SLAP_HEAD]

You're right maubp. I got it in my head that sff_extract could work on FASTA files. My brain must have already left for the weekend.

It certainly would be nice if the SRA allowed you download SFF files, seeing as that is how they are uploaded.

[maubp]

I'm guessing from the typo earlier he will be using MIRA, so either FASTQ or FASTA + QUAL is fine. That is not the problem.

The problem is the reads from the SRA or a raw SFF file go like this:

5' (tag TCAG) (optional MID barcode) (end of fragment) (paired linker) (start of fragment) 3'

They need to be split, giving the start fragment as the "forward read" and the reverse complement of the end fragment as the "reverse read" (to mimic Sanger or Illumina paired end reads). If you have the SFF file, then sff_extract does all that for you.

One option would be to extend sff_extract to take FASTQ input files, that might not be too hard. It is open source.

[kmcarr]

Or you could convert the FASTQ file to FASTA format sequence and qual files. The fq_all2std.pl script (originally part of the MAQ distribution) can handle this conversion using the 'std2qual' subcommand.

Do you need to maintain the data in FASTQ format for downstream analysis or will FASTA be acceptable?

[maubp]

sff_extract uses MIRA to identify the linker sequence in the read.

No, sff_extract uses SSAHA2 to split paired end 454 reads

I guess you could write a python script based on what sff_extract does to split FASTQ files containing paired end Roche 454 data.

[sjackman]

Hi,

I have a FASTQ file downloaded from SRA of paired 454 reads:

http://www.ncbi.nlm.nih.gov/sra/SRX022096?report=full

Each paired read is a single FASTQ record, including the linker sequence somewhere in the middle. What software can identify the linker and split the read into two FASTQ records? I've used sff_extract for SFF files, but the author has confirmed that it cannot handle FASTQ files.

Thanks,
Shaun