[Genome] Not all identities returned by blat

[Brooke Rhead]

Hello Johanne,

I tried BLATing your example sequence against sacCer1 using different 
parameters.  I used -minIdentity=100 to get only matches with 100% 
identity.  Strangely, using the default parameters, I got 19 full-length 
matches, one more than you found.  I am using Standalone BLAT v. 33x5.

I got more 100% identity results by lowering the tileSize.  With 
-tileSize=6 I found 22 full-length, 100% hits.  With tileSize=7, I found 
20, and with tileSize of 8 to 11, I found 19 hits.

I should also point out that with web-based blat on our site (at 
http://genome.ucsc.edu/cgi-bin/hgBlat? ), your example query gives 20 
100% identity, full-length matches.  Of course, some of the results have 
a span greater than 46 (the length of the query).  We have advice on 
replicating web-based BLAT results here:

http://genome.ucsc.edu/FAQ/FAQblat#blat5

(actually, this title is somewhat misleading . . . these are 
instructions on replicating web-based BLAT using gfClient and gfServer, 
not blat.  You still may find them useful.)

Also, if by "perfect matches" you mean something other than the way I 
interpreted your meaning, please let us know, and we will try to advise 
you on how to alter the parameters.

-- 
Brooke Rhead
UCSC Genome Bioinformatics Group


Duhaime Johanne wrote:
> Hello
>  
> I am trying to find all perfect matches of 45-55 oligos on the
> Saccharomyces cerevisiae.
>  
> I would like to use blat instead of megablast because of the speed since
> I  have a lot of oligos.
>  
> A blast with a certain sequence will return 20 matches 100% identity on
> the full oligo lenght but with blat I get 18. And this happens for quite
> a lot of sequences. I always get less matches with blat. 
>  
> I am struggling with the parameters but I cannot get the same results.
>  
> Ex: TAGTCGCACTAGTCCTGACGTTGATGCTGGCAGTGGTAGTAGCACT  blast=20 blat=18
>  
> I am using the following : 
>  
> /apps/programs/bin/blat x11.seq ../../../sgd/all.fsa -out=blast8
> x11.blat
>  
> I have tried to change tileSize an stepSIze without success. 
> The matches for the above sequence that do not appear in the blat result
> are in the neighborhood of another match. 
>  
> Database, blast and  blat are installed locally. 
> The megablast command used: 
> /apps/programs/sources/blast-2.2.15/bin/megablast -d ../../../sgd/all -i
> x11.seq -o x11.blast -m 8 -F f -p 0
>  
> Is there a parameter that will conciliate the result or say differently
> how can I have all the perfect matches with blat?
>  
> Thank you very much in advance
>  
>  
> Johanne Duhaime
>  
> _______________________________________________
> Genome maillist  -  Genome at soe.ucsc.edu
> http://www.soe.ucsc.edu/mailman/listinfo/genome