[DAS] Retrieving feature information

Matthew Pocock matthew_pocock@yahoo.co.uk
Thu, 01 Aug 2002 16:21:55 +0100


Hi Oliver,

This is one of the big things we discussed while designing DAS. There 
are a number of reasons why it is realy technicaly hard to have the 
servers doing the feature projection. BioJava and BioPerl both provide 
library code to do the client-side feature projection for you. Some of 
the reasons it's bad for the server to project are:

* there are more than one way to assemble a bit of genome and the 'best' 
assembly can change over time. Servers currently don't need to know 
anything about the assembly. To project features you do need the assembly.

* It is much harder to efficiently handle queries if you are projecting 
features as the server must project /every/ feature into the request 
co-ordinates to check for overlaps. This is not an issue with servers 
that just return info for one chunk of one level of the assembly when 
requested.

It's good because:

* Simple-stupid client code would be much easier to write

What language are you in? Would you use an open-source stripped down 
client if it was available and easy to use?

Matthew

Oliver Lyttelton wrote:
> 
> 
> Dear Developers,
>    Please consider this plea, it is heartfelt.
> 
> When I first heard about DAS, I was in a lecture at Imperial College on 
> distributed Bioinformatics. The concept seemed perfect, the idea of 
> allowing anyone with annotations to publish them in a standard format, 
> which a single genome browser could interpret, creating a virtual free 
> market on the internet, and thus improving the quality of annotation 
> information for everyone.
> 
> Since I have been using the system, I have come across one massive 
> problem with it, and that is the problem that there is no server-side 
> conversion of coordinate systems during feature requests.
> 
> I understand the reason for wanting to store features at the lowest (and 
> therefore most stable) level of granularity. However, I fail to see why 
> this prevents real-time server-side conversion.
> 
> My basic premise is that a "dumb user" such as myself, ought to be able 
> to query (programatically) a DAS annotation server in the coordinate 
> system of their choice, and retrieve a list of all annotations stored on 
> that server, in the same coordinate system. At the moment the amount of 
> conversion required is prohibitive.
> 
> The task for the server is to translate the request location into 
> chromosome coordinates, and then recurse down through the hierarchy 
> picking up feature info at each level and translating it all into the 
> client coordinate system. This strikes me as a server-side type job not 
> client side, and if someone wrote the code once, it would save every 
> client side program from having to do the same job.
> 
> To prevent queries returning too much info you could limit the number of 
> features returned to a specific number...
> 
> Let me know what you think,
> 
> Best regards to everyone,
> 
> Oliver Lyttelton (MSc Project Imperial College)
> 
> _________________________________________________________________
> Send and receive Hotmail on your mobile device: http://mobile.msn.com
> 
> _______________________________________________
> DAS mailing list
> DAS@biodas.org
> http://biodas.org/mailman/listinfo/das
> 




__________________________________________________
Do You Yahoo!?
Everything you'll ever need on one web page
from News and Sport to Email and Music Charts
http://uk.my.yahoo.com