Get one hit per specie in BLASTp

Question

[Agenor Neto]

Hi everyone! I have searched this answer but it seems a problem that is not yet solved: is there any way to perform BLAST and get just one hit per specie (taxid)?

Right now, I am performing BLASTp locally and it is working fine but, for my current purpose, isoforms are not what I want. Instead, I would like to retrive only the canonical form (and I am aware that this is not an information available for all the proteins). I would like to know if there is some solution, a built program or something I could programmatically do.

[GenoMax]

You will invariably need to post-process/parse your blast output (will be easier if you are using one of the tab delimited formats) to get this type of specific output.

Answer 1

[brian.fristensky]

I'm not certain what you mean by "canonical form", but I'll assume for discussion that you want 1 orthologue from a database such as Uniprot, to represent each species. In the BIRCH system, BLAST search output pops up in several windows, one of which is a BioLegato spreadsheet that lets you sort output in a spreadsheet form.

In the example, you could sort first by TaxID for species, and secondarily for a characteristic such as alignment length or E-value. The top one for each species would therefore be the best hit, longest alignment etc. for that species. Whichever hits you select could then be retrieved from NCBI to a file or a new BioLegato object for subsequent analysis. This process is outlined in the tutorial Searching Local Sequence Databases Using BLAST.

[Agenor Neto]

That's exactly the conclusion I got after reading the BLAST documentation: I am going to try to filter my results using features as e-value, percentage identity etc. For those who are trying the same thing here is a link of parameters which you can define using command line BLAST. Something that helped me to idealize this rationale was also the rationale used in RefSeq_select. It is used for transcripts and only has selected transcripts for human, mouse and prokaryotes but follows the same purpose we are discussing here. That said, when I get this task done, I will post here if this solution worked. Thanks!