[DAS] Multiple methods for a feature

[Andy Jenkinson]

Hi Leyla,

Quite simply, I imagine this is just unique to UniProt. Most peoples' DAS sources are for data generated from a single source, i.e. primary data.

But really I think your problem is that you are wanting to use the method to describe how the data behind your annotations is derived (e.g. an experiment, or publication) NOT how the annotation itself was derived (i.e. inferred by a curator). The semantics are a little fuzzy here, but if you were being pedantic about it you should consider the accuracy of this - is your annotation actually fully expressed by each of the evidence codes you are using?

For a contrived example, a curator comes to a conclusion that protein X is involved in a pathway A via these pieces of evidence:
Experiment 1 says protein X interacts with protein Y.
Experiment 2 says protein X interacts with protein Z.
Paper 3 says proteins Y and Z are part of pathway A.

So in this case, would you use the evidence codes of "inferred by experiment" and "inferred from publication", or whatever they are called, to support an annotation of "protein X is involved in pathway B"? These are not really several methods of producing the annotation - the actual method was the curator's ability to combine the conclusions of the other methods. So would you use "inferred by curator" instead?

Cheers,
Andy

On 17 Oct 2011, at 13:43, Leyla Garcia wrote:

> Hi all,
> 
> At UniProt, we can have more than one method for a feature. It could be the same method but from different sources, for instance "Experimental evidence ECO:0000006" could refer to experiments from Ensembl and InterPro, or totally different methods, for instance "Experimental evidence ECO:0000006" and "Sequence similarity ECO:0000044".
> 
> Currently in DAS 1.6, it is only possible to have one method for each feature. Why? Is there any other DAS server facing a similar situation as UniProt? If there is, how have you solved this problem?
> 
> Cheers,
> 
> Leyla
> _______________________________________________
> DAS mailing list
> DAS at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/das