On Wednesday 10 March 2010 23:38:58 Jim Kent wrote:
> Hmm, 32 base pair translated sequences is pushing the envelope on blat
> a little, but it could work!
You are right.
I hope it would work; I tried to map these sequences to NR with Blastx but
without reliable results. If anybody an alternative way to performe this task,
please tell it to me.
Another approach could be to assemble these sequences till we get contigs
100-128bp long, then stop the assembling process and then compare these
contigs with NR using BlastX. Since these are metagenomic reads, I want to
assemble them as least as possible because there is a high risk of creating
chimeric contigs.
But my colleagues have just CLCBio and it cannot stop the assembling process.
Thank all of you for your suggestions!
Fabio
>
> On Mar 10, 2010, at 5:19 PM, Galt Barber wrote:
> > Jim Kent informs me that you can do it this way:
> >
> > For the command line you can achieve this just by swapping target and
> > query.
> >
> > Your 32bp sequences would become tiny targets,
> > you would need -t=dnax -q=prot.
> >
> > Happy BLAT-TING!
> >
> > -Galt
> >
> > Galt Barber wrote:
> >> Hi, Fabio!
> >>
> >> Here are the type combinations supported by blat from blat.c:
> >>
> >> if (bothSimpleNuc || bothSimpleProt)
> >> {
> >> [...]
> >> }
> >> else if (tType == gftDnaX && qType == gftProt)
> >> {
> >> [...]
> >> }
> >> else if (tType == gftDnaX && (qType == gftDnaX || qType == gftRnaX))
> >> {
> >> [...]
> >> }
> >> else
> >> {
> >> errAbort("Unrecognized combination of target and query types\n");
> >>
> >>
> >>
> >> What you would love to see and what is missing is this line:
> >>
> >> else if (tType == gftProt && (qType == gftDnaX || qType == gftRnaX))
> >> {
> >> [...]
> >> }
> >> But this combination is not supported by BLAT.
> >> Perhaps Jim Kent felt that BLAST already did this well enough
> >> and did not choose to support this.
> >>
> >> You could try converting your reads into 3 frames of protein
> >> and then using blat -prot.
> >>
> >> By the way, if you are specifying -out-blast8 then you wouldn't
> >> typically give your output file name the extension .psl.
> >>
> >> The results you got when using -prot on a dna query
> >> are invalid. Do not use them.
> >>
> >> -Galt
> >>
> >> Fabio Gori wrote:
> >>> Hi,
> >>>
> >>> Sice my reads are just 32bp long, I am trying to use blat instead
> >>> of blastx to
> >>> map them into proteins.
> >>> I am trying to run
> >>> blat nr.fa seqsaglobus.fa ris2.psl -t=prot -q=dnax -tileSize=8 -
> >>> stepSize=3 -
> >>> fine -repMatch=1000000 -out=blast8
> >>>
> >>> But I got the message:
> >>> d and q must both be either protein or dna
> >>>
> >>> I tried with
> >>> -t=prot -q=prot
> >>> and I get some results, but it should not work because
> >>> seqsaglobus.fa is made
> >>> by nucleotides.
> >>>
> >>> Can you explain me what happens?
> >>>
> >>> Thank you,
> >>>
> >>> Fabio
> >>
> >> _______________________________________________
> >> Genome maillist - [email protected]
> >> https://lists.soe.ucsc.edu/mailman/listinfo/genome
> >
> > _______________________________________________
> > Genome maillist - [email protected]
> > https://lists.soe.ucsc.edu/mailman/listinfo/genome
--
F. Gori, PhD student
Intelligent Systems
ICIS (Institute for Computing and Information Sciences)
Radboud University Nijmegen
Post Address:
Intelligent Systems
Postbus 9010
6500 GL Nijmegen
The Netherlands
Visiting Address:
Room HG02.517
Faculty of Science
Heyendaalseweg 135
6525 AJ Nijmegen
Tel.: +31 (0)24 36 52703
E-mail: [email protected]
Home Page: http://www.cs.ru.nl/~gori/
_______________________________________________
Genome maillist - [email protected]
https://lists.soe.ucsc.edu/mailman/listinfo/genome
