[Bioperl-l] Building consensus sequence from ESTs

[Elia Stupka]

> Fine. I've been looking through the pages at biopipe.org. I've seen that
> you plan to use GMOD or some other interfases to look at the results of
> the pipeline. What is your idea regarding StackPack and the clustering
> of ESTs? Perhaps you only want i) a non-redundant dataset to ease the
> mapping of transcript data to the genome. If this is the case, perhaps
> you don't need any display at all (like the one that StackPack provides
> through its web interface).

We already have StackPack running with its viewer. The integrated viewer
is a neat solution for looking at the actual results. But the idea is to
also be able to integrate StackPack clustering as "just a step in your
pipeline", whether you display things in GMOD or ENSEMBL.

> But if you want to create an EST analysis pipeline, perhaps you will be
> reproducing much of what stackpack already does. 

We definitely will not want to do any work that StackPack does already,
the idea is just to wrap first all of it and then some components of it
just to enable parallelisation.

> The main benefit, being associating other kinds of data to the
> clusters (similarity mainly, but also expression data from microarrays
> perhaps). Is this your idea?

Absolutely, by writing wrappers for it we can make it part of a bigger
workflow such for example correlation with microarray data, or using
consensuses to build genes.

> Thanks for keeping me informed. Although we don't have a compute farm,
> we could also benefit from the pipeline.

StackPack actually runs best not on a farm but on an SMP machine (e.g. Sun
8-way machines, Power4,etc.) because it scales proportionally with number
of CPUs running together.

The pipeline doesn't need a cluster, it just helps you to manage work on
one if you have it, so don't worry :)

Elia

********************************
* http://www.fugu-sg.org/~elia *
* tel:    +65 6874 1467        *
* mobile: +65 9030 7613        *
* fax:    +65 6779 1117        *
********************************