[DAS] LDAS vs Dazzle

jfreeman jfreeman@variagenics.com
Mon, 07 Jan 2002 11:22:16 -0500


Ewan Birney wrote:
> On Fri, 4 Jan 2002, jfreeman wrote:
> > Hello all,
> >
> > Thanks for the quick response!
> >
> > Ewan Birney wrote:
> > >
> > > Just to make sure people don't think something weird is going on:
> > >
> > > (a) Ensembl's "assembly" is (currently) a particular UCSC golden path and
> > > soon will be one of the NCBI golden paths. Ensembl doesn't make assemblies
> > > - we just choose one to use.
> >
> > I never meant to imply that Ensembl's choice of any of the golden paths
> > in the past or present or future was or is or will be suspect, from my
> > last note.  I was, badly, making the point that if you decide to map
> > your data onto any one of them the cost of doing so leads you to want to
> > pick one, as Ensembl does, rather than map to them all.  That is the
> > source for a natural lock-in for anyone doing the mapping, if you would
> > like enjoy the utility of the Ensembl package then you go with what
> > Ensembl chose when you do your mapping.  The choice of a particular
> > golden path is Ensembl's prerogative as you mention, but the rationale
> > for the change is not clearly understood, given your cost of remapping
> > your data, and my cost of doing the same.  If you have the time to
> > educate me, I would like to understand what leads to the Ensembl group
> > choose a particular golden path, and this will help me explain the value
> > of the system, given the remapping costs, internally, and might be of
> > some use to others on this list.
> <grin>
> I think this is the sort of conversation one needs to have over beers.

Step one in Extreme Conversing? :)  

>   - We have to move golden paths because we want to encorporate new data
> (remember - the human genome is being finished all the time).
>   - currently there are two golden paths per freeze date of the data (NCBI
> and UCSC) but they are both converging to the same set of underlying data
> inside the sequencing centres for each chromosome. From here on in for
> human assemblies it is really a rather boring data tracking exercise (but
> of course very necessary).
>   - We've spent the last 9 months setting up this data tracking system in
> a better way (for the people in the know these are the "TPF files"). To
> help everyone work together we are now focusing on just one assembly
> series - NCBI - with the agreement of UCSC (UCSC is going to keep NCBI
> honest and move onto bigger/better things).
>    Does this make sense?

Excellent.  It makes sense to me and thanks.  The DAZZLE system issues I
have talked about are small in comparison to that and will work
themselves out over time.

Thanks again,