Hello again,

Here are the comments from one of our Blat developers:

---start----
There is no excellent statistics theory for alignments
with gaps.

In practice what people do is use pslReps
and pslCDnaFilter programs to filter for the alignment
properties they want such as coverage, percent identity,
etc.

One may be looking for the single
best alignment, or the alignments that are nearly
as good as the best alignment.

Even if one had a statistical e-value, the
cutoff one uses is sometimes a little arbitrary,
and other post-filtering of alignments may
be required based on other criteria.
-----end----

The programs mentioned above "pslReps and pslCDnaFilter" are a part of 
our code tree. To use them, you will need to obtain the code tree and 
compile the programs as necessary. Here are links to web page listing 
some of our common utilities and instructions for obtaining the code tree.
http://genomewiki.cse.ucsc.edu/index.php/Kent_source_utilities
http://genome.ucsc.edu/FAQ/FAQdownloads#download27

We hope these help,
Jennifer Jackson
UCSC Genome Bioinformatics Group


Mittal, Vinay K wrote:
> Thanks for your prompt reply.
> Thats a good idea to keep the BLAT score as an alignment metric. But what 
> should be the cutoff score to call an alignment statistically significant and 
> not just happen to be there by chance? Because usually we take E-value less 
> than 0.01 as significant. Is there any kind of such cutoff for BLAT score?
>
> Thanks once again.
>
> V.
_______________________________________________
Genome maillist  -  [email protected]
http://www.soe.ucsc.edu/mailman/listinfo/genome

Reply via email to