[Gusdev-gusdev] Re: Protein properties and protein features in GUS

[Arnaud K. <ax...@sa...>]

Chris

I realise I did some propositions about new protein features but I've 
never done their formalisation in SQL statements.
Shall I do that to prepare their incorporation into GUS ?

cheers
Arnaud

Arnaud Kerhornou wrote:

>  Hi everyone
>
> I would like to report a new table, ProteinProperty and new views on 
> the top of AAFeatureImp table for protein features such as domains.
>
> * Protein properties :
>
> There are 4 protein properties :
>   * Isoelectric point (1),
>   * Molecular mass (2),
>   * Charge (3),
>   * Average residue mass (4).
>
> The 3 first ones may have several values as they can be characterized 
> experimentally.
>
> From a design point of view, we can have a unique ProteinProperty 
> table or a view foreach proterty (a ProteinPropertyImp table and three 
> views: IsoElectricPointProperty, MolecularMassProperty and 
> ChargeProperty).
> The number of properties may not changed in the future so I may be 
> simpler to create a unique ProteinProperty table.
>
> Specification => A property would behave like a feature, ie :
>   * it is attached to a sequence modulo the fact it doesn't have a 
> location within it,
>   * it can be supported by evidences such as an experiment, published 
> or from a personal communication.
>   * have external db refs.
>
> ProteinProperty table:
>   * protein_property_id : number
>   * property_name : varchar2(50)
>   * property_value : number (5)
>   * & stuff common to any GUS table: modification_date ...
>
> The 4th property, average residue mass, could be an extra attribute in 
> the proteinSequence or TranslatedAASequence view.
>
> ******************
> * Protein Features :
>
> Features attached to a protein sequence.
>
> The new features objects are:
>    (1) Signal Peptide Feature :
> It's already a view in GUS, but we will store curated data, such as 
> targetting information.
>
>    (2) Domains:
> It can be:
>    * a Leucine Zipper domain,
>    * a coiled-coil domain,
>    * a Pfam, Smart or Prosite domain.
>
> DomainFeature view:
>    * aa_feature_id : number (10),
>    * aa_sequence_id : number (10),
>    * name : varchar2 (50),
>    * description : varchar2 (100),
>    * score : number (4)
>    * e_value : number (10),
> + external database link entries and a location object.
>
>    (3) Transmenbrane domain feature:
> Question : PlasmoDB web site shows hydrophobicity graphics, where is 
> it stored in GUS ?
>
>    (4) Post-translational modification feature:
>    * type : varchar2 (50) (e.g. glycosylation, phosphorylation ...)
>    * modified_by : use of the Interaction table ?
>    * Coordinates of the phosphorylation site in a AALocation object.
>
>    (5) Repeat Features, should be the same design that at the DNA level :
>        * RepeatRegionFeature as a set of RepeatUnitFeatures,
>        * RepeatUnitFeature, with the consensus sequence, name and size
>        * RepeatType table
>
> Another question : What about 2D structures (beta-sheet and 
> alpha-helice) in GUS ?
>
> Let me know if you have any comments. I'll send another email for 
> extra features at the DNA/RNA level.
>
> Cheers
> Arnaud
>