[Genome] blat: minIdentity option not working?

[Galt Barber]

Hi, Adnan!

What's happening here is that blast only gives you the alignments of each
exon individually.  You have to do some post-processing to chain all these
together into a full alignment.  BLAT however saves you all that work by
doing the chaining together for you into a single final alignment, in psl
format by default.  The -minIdentity=90 refers to the identity of that
full alignment.  When you ask it to output as blast8, it has to break the
exons back apart to satisfy blast's simpler output structure.  So those
exon identities will not individually be 90%, but collectively they are.

Notice that in your example the identities that are below 90% are not very
far below it. There's nothing like 50% or 30%.  Most of them are slightly
over or slightly under 90%, and the collective average will be >= 90% as
the blat help for minIdentity states.

Take care and happy blatting!

-Galt

> Thanks for responding. Here's an example where blat produces fragments
> with less than 90% identity. I aligned NM_133647 (a Refseq mRNA
> sequence) with human chromosome Y (UCSC v. HG18), with default options
> except for the output (and the use off ooc):
>
> blat hg18.chrY.fa NM_133647 -out=blast8 -ooc=hg18.chrY.fa.ooc
> NM_133647.vs.chrY
>
> Here is the beginning of the output:
>
> NM_133647       chrY    87.59   274     34      0       4967    5240
> 1540705 1540432 1.0e-121        433.0
> NM_133647       chrY    92.13   216     16      1       5688    5902
> 1772513 1772298 2.4e-100        362.0
> NM_133647       chrY    91.78   219     16      2       5688    5905
> 2779105 2778888 3.9e-100        362.0
> NM_133647       chrY    90.87   219     16      3       5629    5844
> 2616703 2616486 5.1e-98 355.0
> NM_133647       chrY    85.06   241     33      1       5022    5259
> 2298461 2298221 1.8e-96 350.0
> NM_133647       chrY    93.27   208     11      3       5689    5894
> 292466  292672  1.2e-95 347.0
> NM_133647       chrY    90.55   201     18      1       5688    5887
> 1815749 1815949 3.6e-92 335.0
> NM_133647       chrY    86.64   217     29      0       5022    5238
> 714027  714243  5.8e-92 335.0
> NM_133647       chrY    88.52   209     24      0       5030    5238
> 2670301 2670093 9.6e-91 331.0
> NM_133647       chrY    90.36   197     17      1       6120    6316
> 2227040 2226846 5.9e-88 321.0
> NM_133647       chrY    89.85   197     18      1       6120    6316
> 5207455 5207261 1.3e-87 320.0
> NM_133647       chrY    89.11   202     15      2       5727    5922
> 3454521 3454321 1.3e-86 317.0
> NM_133647       chrY    89.34   197     19      1       6120    6316
> 5251242 5251436 1.4e-85 313.0
>
> I get the same output if I specify -minIdentity=95 or -minIdentity=90. I
> chose this example at random, and I get matches like this (less than the
> % identity specified) routinely (basically for every mRNA-chromosome
pair).
>
> FYI, I'm running blat v. 31x1 on a Linux machine.
>
> Thanks.
>
> aDNAn


On Mon, 16 Oct 2006, Adnan Derti wrote:

> Hi.
>
> I'm using blat (v. 31x1) locally to align mouse and human mRNAs to their
> respective genomes. Blat produces aligned segments with <90% identity
> whether I use the default options or specify -minIdentity=90. I use
> -out=blast8 and haven't checked whether other output options produce the
> same results. Am I misunderstanding the way -minIdentity should work?
>
> Thanks.
>
> Adnan Derti
> _______________________________________________
> Genome maillist  -  Genome at soe.ucsc.edu
> http://www.soe.ucsc.edu/mailman/listinfo/genome
>