[Bioperl-l] Re: Added Phil Green's Smith-Waterman to my module

[Yee Man]

> No, I did -m 9c which is like -d 1 except it doesn't spam my screen with
> output.
> 
> And when I run the command line you gave, I get the same 7 seconds of
> alignment calculation, and still 40 with yours.
> 

ok.

> 
> I'll look into it soon, but I encourage you to "double check":
> 
> % /usr/bin/time perl -I./blib/lib -I./blib/arch align.pl t1.fa t2.fa > /dev/null
>        41.43 real        40.87 user         0.21 sys
> 
> [ this includes about 1.5 seconds of Perl "overhead" ]
> 
> % /usr/bin/time ../facurr/ssearch34 -3 -n -q -H -d 1 -b 1 -f 3 -g 1 -r +3/-1 t1.fa t2.fa > /dev/null
>        10.43 real        10.40 user         0.03 sys
> 
> [ This includes about 2.5 seconds of "search" time ]

I repeated the exact command line at my machine and it still yields a slow
run. What kind of time do  you get when you run it with score only? Mine
is 12sec for score only and 70sec for score and alignment. Theoretically
speaking, the alignment phase with Hirschberg's algorithm should take
several folds the time you calculate the score (My guess is about 2 times 
but I guess I need to implement one to be sure). So unless you only spend
2 or 3 sec to calulate the score, you shouldn't be able to get the
alignment in 10sec.

Also, my ssearch34 is compiled with gcc with -g and without -O and other
options. Will that be the cause? Can you compile a ssearch34 with gcc with
-g as the only option? Compiler and compiler options can sometimes give
you some dramatically different speeds.

Yee Man

> 
> -Aaron
> 
> -- 
>  Aaron J Mackey
>  Pearson Laboratory
>  University of Virginia
>  (434) 924-2821
>  amackey at virginia.edu
> 
>