[Bioperl-l] Error parsing blast results with blasttable

[michael watson (IAH-C)]

Hi

Having done some more tests with this:

$hit->start()

Actually returns a string which is the concatenation of query start and
subject end!  (btw I'm using the "-m 8" option) - surely this isn't the
desired option????

If I change it to:

$hit->start('query')

Then I get the correct start back, but I still get the stack trace
error.

The two co-ordinate sets which cause the problem (3264-3268 and
3252-3268) are on adjacent lines in the file (3252-3268 is the next line
after 3264-3268) and are to the SAME subject, ie they are two HSPs of
the same hit (in theory) but they are to two VERY different parts of the
query.

I'm guessing the way blasttable handles multiple HSPs is causing the
trouble.

Mick

-----Original Message-----
From: Marc Logghe [mailto:Marc.Logghe at devgen.com] 
Sent: 07 January 2005 10:41
To: michael watson (IAH-C); Bioperl List
Subject: RE: [Bioperl-l] Error parsing blast results with blasttable


> while (my $result = $searchio->next_result) {
>         while(my $hit = $result->next_hit) {
>                 my $start  = $hit->start;
> 
> And it is that call to $hit->start that sets off the whole trace.
> 
> Any ideas?


Hi Mick,
Have you tried one of these ?:

my $start = $hit->start('sbjct');  # or 'query' or 'hit'. Latter is same
as 'sbjct'

or 

my $start = $hit->hsp->start('sbjct');


I think in all cases it defaults to 'query'. So it should not crash but
give you the start position of the query. I am afraid I can't explain
the crash, sorry.

Marc