PROSITE motif search on genomic data?

[Paddywhacker]

Is there a tool that gives you pBLAST type searching with a PROSITE type protein specification?

Thanks,
Paddy

[Astarte]

Most motif search tools only accept protein sequences, I have not yet found a tool that accepts genomic DNA. The best approach would be to translate your sequence into all possible reading frames and submit these to prosite. If you know a bit of Perl, R, Matlab, this is something you can easily automate.

[Paddywhacker]

Yes, that is what I want, and that is what I have done in the past. But doing so requires downloading the sequence of the genome. It would be very useful if the online BLAST engines could be upgraded to do regular expression searching.

[Astarte]

it's not really clear to me what exactly you want to do. Blast engines accept IUPAC code, and although this is not as flexible as reg ex, it is usually all you need in the genome world.
Do you want to do a motif search on a whole genome? Or are you looking for a specific motif in a genome?

[Paddywhacker]

it's not really clear to me what exactly you want to do. Blast engines accept IUPAC code, and although this is not as flexible as reg ex, it is usually all you need in the genome world.
Do you want to do a motif search on a whole genome? Or are you looking for a specific motif in a genome?

I'm defining motifs from alignments between my model and the few number of annotated genomes that I can find and then trying to find matches in whatever raw genomic scaffolds that I can find online. The modifs are protein based with ambiguities and variable filler lengths (typical PROSITE), because that is where the conservation can be seen, but the scaffolds are DNA.

[Astarte]

The IUPAC code will take care of the ambiguities, but the variable filler length is a problem. you can try with a low score for the creation of a gap. This will hopefully give you a good first selection. But I'm afraid from here on, you will have to download this first selection and do a more stringent matching with a reg ex in matlab or R.

Good luck!

[Paddywhacker]

Thanks for your input, but the IUPAC codes don't have a mechanism to restrict the ambiguities to a a specific set. 'X' stands for any amino acid residue when you might want a choice between say, Phe and Try.

It looks as if I have to carry on what I've been doing. Download all scaffolds. Convert each scaffold into the six different protein frames. Scan each repeatedly for motif matches, increasing the permitted error count each time until I get a match. And examine the genomic neighbourhood of the match to see if it isn't just a garbage hit.

[Astarte]

Can't you translate your protein motif into a DNA sequence, for nucleotides, each possible ambiguity is met in the IUPAC code. Then use this to do a regular nblast?