[Bioperl-pipeline] UPDATE action

Kiran kiran@fugu-sg.org
Fri, 10 Jan 2003 01:37:34 +0800


Hi Jeremy,

> The pipeline I am working on right now uses the "UPDATE" action. I found
> that when this action is used with later analyses (from the third analysis
> onward), it doesn't seem to work (I think this is because the second
> analysis jobs are created in the actual InputCreate step). I added
> dependency checking code to the section of PipelineManager.pl that builds
> new jobs. The code looks like this:
>
>
>     foreach my $job (@completed_jobs) {
>         my ($new_jobs) = &create_new_job($job);
>         if(scalar(@{$new_jobs})){
>             print STDERR "Creating ".scalar(@{$new_jobs})." jobs\n";
>         }
>         foreach my $new_job (@{$new_jobs}){
>             my $job_depend =
>                 $ruleAdaptor->check_dependency_by_job($new_job,@rules);
>             $new_job->dependency($job_depend);
>
> The last two lines are the addition.

The dependency check is to tell the pipeline whether to store the outputids
of the job in the pipeline database or not. (These outputids are needed to
be stored for certain actions like UPDATE and WAITFORALL_AND_UPDATE as they
would serve as the inputids for the jobs of the next analysis)
If the UPDATE action is used, for every outputid returned by the analysis's
output iohandler, a new job of the next analysis is created with this
outputid as the input.
The likely reason that i guess is that your 2nd analysis's outout iohandler
(if you specified one) doesnot return any outputids, so no jobs will be
created for the next analysis.
(Please note : This feature will be obfuscated soon as we have observed that
with correct design of pipeline (its input/output storing and fetching) you
may never need this in practice)
Also, The lines you have added would have no effect on the pipeline as there
is already a dependency check before that in the pipeline manager.

anycase if you can share the Xml file you are using to run the pipeline and
a short description of what you are trying to do, we can understand your
problem more clearly and guide you correctly.. also we are trying to come up
with decent documentation soon..

Kiran









>
> Jeremy
>
>
> _______________________________________________
> bioperl-pipeline mailing list
> bioperl-pipeline@bioperl.org
> http://bioperl.org/mailman/listinfo/bioperl-pipeline
>
>