UCSC Genome Browser discussion forum ()

Mbandi S.K | 14 Jun 2012 22:23

blat: sim4 Vs psl form

Dear ALL;

Firstly, I'm happy to join this mailing list. I do not know if this group
is the right place for my question. Kindly bear with me if my question is
trivial or has been dealt with already. I have recently settled on BLAT v.
34 for a portion of my project to screen for EST(cDNA) that well aligned
to my reference sequence. However, I find it hard to understand the
effects of -minIdentity and -fastMap on the output.

I also noticed that just changing the output format, affects the the
reports in the output file. More ESTs are reported in sim4 format than in
psl format. I want to write a parser to calculate coverage, identity etc
in other for me to build a filtering matrix. attached here are two test
files:query.fa and target.fa. I'm aware -fastMap is for DNA-DNA, but just
for test purposes, I ran:
blat target.fa -t=dna query.fa -q=dna -out=psl -minIdentity=100 -fastMap
-dots=1 test.psl
and
blat target.fa -t=dna query.fa -q=dna -out=psl -fastMap -dots=1 test.psl

However in the first instance; I do not find hits which I expected even
though default -minIdentity is 90 which is less stringent to 100. When
out=sim4 is used, the hits are totally different. Query.fa contains
mutated and unmodified versions of seq1 from target.fa file.

Has anyone experience strange results like this? Which output is better
from experience? I will appreciate clarity in this regard.

Many thanks,

(Continue reading)

headers

Luvina Guruvadoo | 18 Jun 2012 22:26

Re: blat: sim4 Vs psl form

Hi Mbandi,

Thank you for contacting the mailing list, and yes, this is the correct 
place to ask your question. One of our engineers suggests you use psl 
since it is the native blat format, and not to use fastmap for ests 
which may have introns in them. In addition, you may download our latest 
version of BLAT which contains a few bugfixes and may be useful for your 
purposes. The lastest BLAT is available in compiled form in our 
downloads: 
http://hgdownload.cse.ucsc.edu/downloads.html#source_downloads. We also 
suggest you use pslReps and pslCDnsFilter for filtering psl results.

I hope this information is useful and answers your question. Please 
contact us again at genome <at> soe.ucsc.edu if you have any further questions.

---
Luvina Guruvadoo
UCSC Genome Bioinformatics Group

On 6/14/2012 1:23 PM, Mbandi S.K wrote:
> Dear ALL;
>
> Firstly, I'm happy to join this mailing list. I do not know if this group
> is the right place for my question. Kindly bear with me if my question is
> trivial or has been dealt with already. I have recently settled on BLAT v.
> 34 for a portion of my project to screen for EST(cDNA) that well aligned
> to my reference sequence. However, I find it hard to understand the
> effects of -minIdentity and -fastMap on the output.
>
> I also noticed that just changing the output format, affects the the

(Continue reading)

headers

Mbandi S.K | 18 Jun 2012 23:53

Re: blat: sim4 Vs psl form

Hello Luvina,

Thank you very much. I followed the link provided herein but could not
find blat binaries for 32bit. There is only a liftOver file in linux.i386
directory.

I still observe some strange differences in terms of what is reported in
the sim4 and psl formats for the same set of parameters. In my previous
email, I provided two files: target.fa and query.fa both of which are test
files. If you execute the following:
blat target.fa -t=dna query.fa -q=dna -dots=1 -out=psl test.psl
blat target.fa -t=dna query.fa -q=dna -dots=1 -out=sim4 test.sim4

The outputs test.psl and test.sim4 do not seem to contain the same number
of hits. Suprisingly, .sim4 has more hits than .psl although .psl is the
generic format. I am unable to explain how this is possible. For now, I
using the sim4 format and albeit difficult to parse.

Thank you in advance.

Mbandi
------------------------------------
Universiteit van wes kaapland

> Hi Mbandi
>
> Thank you for contacting the mailing list, and yes, this is the correct
> place to ask your question. One of our engineers suggests you use psl
> since it is the native blat format, and not to use fastmap for ests
> which may have introns in them. In addition, you may download our latest

(Continue reading)

headers

Galt Barber | 20 Jun 2012 00:40

Re: blat: sim4 Vs psl form

Hi, Mbandi!

The minIdentity and minScore filtering
have not yet been implemented for sim4 output -out=sim4.

If you run blat -out=psl -minIdentity=0 -minScore=0
you should see the same alignments as sim4.

The output order of the sim4 alignments is however reversed
compared to psl.

Jim Kent, the author of BLAT, has now been informed of these issues.

These issues have gone unnoticed so long because almost nobody is using
sim4.

-Galt

<mbandi <at> sanbi.ac.za> wrote:

> Hello Luvina,
>
> Thank you very much. I followed the link provided herein but could not
> find blat binaries for 32bit. There is only a liftOver file in linux.i386
> directory.
>
> I still observe some strange differences in terms of what is reported in
> the sim4 and psl formats for the same set of parameters. In my previous
> email, I provided two files: target.fa and query.fa both of which are test
> files. If you execute the following:

(Continue reading)