Dear Christoph,

On 4 November 2010 10:39, Christoph Helma <helma at in-silico.ch> wrote:

> Dear All,
>
> Under POST /dataset I have found the description "entries for existing
> compound/feature pairs will be overwritten, entries for new
> compound/features will be added". I guess that should be moved to PUT
> /dataset/{id} - is this correct?
>

I guess so.


>
> To replace a dataset at an existing URI I would use POST /dataset/{id} -
> shall we add that to the API?
>
>
OK


> If a dataset has been created by an algorithm or model I would add
> ot:hasSource and ot:parameters to the metadata to make dataset creation
> reproducible.


Make sense indeed.

BTW it makes sense as well to extend ot:hasSource (a better name anybody?)
to compounds/conformers as well  - would be useful to find where the
structure has been obtained from, e.g. different database, structure
optimisation/conformer generation algorithm , etc.

Also a related issue was discussed in the Substructure API wiki page - how
do we handle the case where a dataset is generated by an algorithm/model ,
but subsequently modified (by adding features, compounds) ?  Set "read only"
property for such datasets?


>  In theory I should add the same info to the features, but
> this is highly redundant and could use a lot of space for large feature
> sets (eg. from substructure mining). Would it be sufficent to specify
> the datset with ot:hasSource for each feature?
>
>
Not sure if this is sufficient, as features are independent objects and not
necessarily belong to a single dataset.

Best regards,
Nina


> Best regards,
> Christoph
> _______________________________________________
> Development mailing list
> Development at opentox.org
> http://www.opentox.org/mailman/listinfo/development
>