BCL files

[Daniel Standage]

I was discussing short read mapping with one of my colleagues recently and suggested that she may get more of her Illumina reads mapped if she tried GNUMAP. I told her about the support for .prb files and she said she would give it a try.

Turns out that the sequencing facility does not create .prb files any more by default, they create .bcl files. Does GNUMAP support this format? What is the recommended format for best performance with GNUMAP?

--
Daniel S. Standage
Graduate Research Assistant
Bioinformatics and Computational Biology Program
Iowa State University

[Nathan Clement]

Daniel,

I've been doing a little hunting and can't find specs for BCL files anywhere (except for a mention that it's a binary format of some sort).  If you can direct me to either a document or a little better description of what they might look like, it could probably be implemented fairly easily.

The best performance for GNUMAP is some type of prb format.  If the files are fastq (or even fasta), GNUMAP first changes them to prb, then maps them as prb.  The advantage of a prb file is that it not only gives a probability score for the highest, "called base," but will give one for every other nucleotide.  In high confidence reads you generally won't have a problem (when the probability score is close to 99.9%), but when there's low confidence, we can still milk some sensible data out of them.


Nathan

[Daniel Standage]

The sequencing center attached this document in their reply to my colleague.

--
Daniel S. Standage
Graduate Research Assistant
Bioinformatics and Computational Biology Program
Iowa State University

[Nathan Clement]

It looks like bcl files are just binary fastq files.  I was hoping they were more like the prb files Illumina used to create.  

If this would make processing much easier (don't need to create other files, etc), I would be glad to add the functionality.  Just let me know.

Nathan

<UsingBCLConverter_15011195_A.pdf>

[Daniel Standage]

Nathan,

I'd say yes if the .bcl files contained probabilities for each base at each position, but since they don't I wouldn't worry about supporting this format just on our account. Storage space isn't our concern, and since Fastq files are easier to work with, we use those.

I'll just make sure that my colleagues explicitly request prb files (in addition to fastq, bcl, and whatever else) from the sequencing center in the future).

Thanks!

--
Daniel S. Standage
Graduate Research Assistant
Bioinformatics and Computational Biology Program
Iowa State University