Blasting human SOLiD short reads against human genome reveals no sig results

[lanner]

I have converted SOLiD SRA to fasta using fastq-dump.

Then, I used the Galaxy groomer to remove the adapter. After that, I wrote a script to translate "0123." to "ACGTN" in the fasta file.

Then, I tried blasting that fasta file against the human genome, and found no significant results.

Earlier, my SAM alignment (derived from the converted fasta file and a reference .mrna human file) also showed very few hits (only 406 out of 6 million).

So, now the finding from BLAST makes me think that the conversion of the human SOLiD sra file went awry somewhere.

I got my .sra file from here (http://trace.ncbi.nlm.nih.gov/Traces/sra/?run=SRR452364, by downloading the "Run" HTTP file).

Based on the above information, does anyone have any ideas where I went wrong in converting the SOLiD sra file to a format ready for BLAST/SAM?

Many thanks...

[Brian Bushnell]

lanner,

You can't translate colorspace to base-space like that, even though the reads give you the first letter. It can only be done after mapping because of Solid's incredibly high error rate - everything after the first error will be wrong.

Of course, it WOULD work for error-free reads, and there should be a few... so try searching for the absolute highest-quality reads, and you might be able to translate and blast those.

[Chipper]

It would not work even for error free reads since he is not converting from colorspace to basespace.

[lanner]

Thanks Brian and Chipper. I realize now that basespace to colorspace conversion is much more difficult than just a string translation!