Helicos SRF to FASTQ

[ohofmann]

Dear all,



been trying to explore a number of Helicos runs. I have access to the read data both in Helicos' SMS format as well as their implementation of SRF (which does not bode well). Using the Staden iolib package it is straightforward to convert reads to FASTA using the srf2fasta command:

>HE-0243693211002-14-2-2-4250
CAGAGTATAGACCTA

Trying the same with srf2fastq results in:

Zero or greater than one CNF chunks found.

My guess is that the quality information is encoded differently in the Helicos' implementation which would explain why Helisphere also only ships with a '2fasta' option.

Is there *any* way to get FASTQ (or a equivalent expressive format) out of Helicos? I'd love to throw the data at a few standard pipelines that usually start with FASTQ data, and seem to be out of options.

Thanks!

[nilshomer]

Dear all,



been trying to explore a number of Helicos runs. I have access to the read data both in Helicos' SMS format as well as their implementation of SRF (which does not bode well). Using the Staden iolib package it is straightforward to convert reads to FASTA using the srf2fasta command:



Trying the same with srf2fastq results in:



My guess is that the quality information is encoded differently in the Helicos' implementation which would explain why Helisphere also only ships with a '2fasta' option.

Is there *any* way to get FASTQ (or a equivalent expressive format) out of Helicos? I'd love to throw the data at a few standard pipelines that usually start with FASTQ data, and seem to be out of options.

Thanks!

Convert to FASTA and then insert dummy quality values (not the best but it will work).

[ohofmann]

Sorry, should have clarified that -- Helisphere even provides scripts that insert dummy quality values, but since we are trying to get a better handle on error profiles that is of course less than ideal. Can't even do basic filtering by quality, or re-evaluate the quality distribution after trimming with FASTX.

Ah well. Guess there's always room for yet-another-format.

[kmcarr]

Trying the same with srf2fastq results in:

Zero or greater than one CNF chunks found.

You need to use srf2fastq with the '-c' option, which tells the program to use the Calibrated Quality scores in chunk CNF1. Without it you get that oh so unhelpful error message. This also happens with SRF files of Illumina data created using illumina2srf in the SequenceRead 2.0 package.

% srf2fastq -c myFile.srf > myFile.fastq

[ohofmann]

Getting the same error using the `-c` flag; maybe indicative of something being wrong with my SRF files rather than the conversion process. I'll see if I can get this to work on public Helicos data first.

Thanks again!