blastx parameters for CDS finding

[rdu]

I'm thinking to use BLASTX to find CDS (protein coding sequence) regions on genomes. How can I set the parameters to let BLASTX put big penalty on a stop codon found? Any experience or any useful link? Thanks.

[robs]

You might want to take a look at ORF finders instead of using BLASTx for this task. If you could provide more information of what you are trying to do, then maybe we could provide more useful help.

[tonybolger]

I'm thinking to use BLASTX to find CDS (protein coding sequence) regions on genomes. How can I set the parameters to let BLASTX put big penalty on a stop codon found? Any experience or any useful link? Thanks.

In short - Don't

Use Scipio or Exonerate - this is exactly what they're designed for.

[robs]

In short - Don't

Use Scipio or Exonerate - this is exactly what they're designed for.

The problem is that there is not enough information in the question. If he/she is interested in prokaryotic genomes, then Scipio would not be very useful.

Eukaryotic genomes:

http://www.webscipio.org/

Prokaryotic genomes:
http://rast.nmpdr.org/

[tonybolger]

The problem is that there is not enough information in the question. If he/she is interested in prokaryotic genomes, then Scipio would not be very useful.

True on both fronts. Still, regardless of the actual aim, there's probably a better tool out there than 'raw' blastx.

[SES]

The problem is that there is not enough information in the question. If he/she is interested in prokaryotic genomes, then Scipio would not be very useful.

Eukaryotic genomes:

http://www.webscipio.org/

Prokaryotic genomes:
http://rast.nmpdr.org/

Since only a genome sequence was mentioned, maybe running sixpack to find all ORFs, followed by iprscan or a Conserved Domain Database search would be more what is requested. Ideally you would combine multiple sources of evidence, and more information would be helpful.

[rdu]

Sorry, I didnt get a chance to back here.

Actually, I'm doing this. I downloaded 20 different bacteria genomes from NCBI website. These genome gene annotation files are ready there. Then I cut each genomic DNA sequence to short read length of pieces (~100nt each) to simulate the ngs reads. I was intending to use BLASTX and protein COG database to annotate these short reads to check how the short reads inherit the function annotation. Let me know if it's still not clear.

Thanks for all your suggestions. But I just want to stick to one of NCBI provided tools.
Happy new year to everyone.

[jordi]

I carried out something similar with 454 reads from a non-model specie, so I had to be aware to point mutations in stretches of homopolymers that could change the frame translation pattern giving rise a false stop codon. My blast parameters were not so restrictive as I think you want. I used a e-value threshold of 0.001 over repetmasked reads using BLASTN (BLASTX was really unsuccesful).
hope this helps and happy new year!