Hi Pantelis, Surajit,

you should take a look again at the following graphical overview of OT
components:
http://www.opentox.org/dev/apis/api-1.2/OpentToxResources/view 

Note that a feature does not have to be coupled explicitly to a dataset,
as Pantelis already said. In practice however it will be most of the
time. The coupling of features - compounds - datasets is done with the
DataEntry. You can in fact (we are doing this to create prediction
features in model building) POST a new Feature without a dataset.
HasSource in this case points to the algorithm that created the feature.

Furthermore, a feature doesn't have to be coupled to a compound. Only
the values the feature takes.

> To be clear lets consider an example of a Dataset X having 20
> compounds. I create a feature which will be assigned to Dataset X
> (that is Dataset X will have a property which is this feature). 

Are you talking about features (e.g. substructural features) that are
dependent on the dataset? So the dataset can be considered a parameter
of the feature generation [descriptor calculation] algorithm? In this
case the service will be very similar to our FTM service. I am a
tutorial/documentation document which will be ready by tomorrow, I
guess. This document should make it easier for you. 


best regards,
Tobias
-- 
Dipl.-Bioinf. Tobias Girschick

Technische Universität München
Institut für Informatik
Lehrstuhl I12 - Bioinformatik
Bolzmannstr. 3
85748 Garching b. München, Germany

Room: MI 01.09.042
Phone: +49 (89) 289-18002
Email: tobias.girschick at in.tum.de
Web: http://wwwkramer.in.tum.de/girschick