Re: More on NWChem and Chempound

[Peter Murray-Rust]

I have copied this to the list

On Thu, Nov 10, 2011 at 7:52 PM, Jorge Estrada <jorge....@unizar.es> wrote:

We are still studying how to add several data fields to Chempound.

We would like to see the following fields:

compchem:package
compchem:version
compchem:title
formula
n:Total energy/n:e_scf
n:wavefunction
n:charge

I'm going through the DepositNWChem code, and I see it generates the RDF that Chempound uses, but I still don't know how it generates the RDF, nor how to add our desired fields from the NWChem CML file.

How is that RDF data generated? How is that RDF data processed by the Chempound server and stored?

How are the RDF data elements selected to appear on Chempound pages?

Jorge

--
Peter Murray-Rust
Reader in Molecular Informatics
Unilever Centre, Dep. Of Chemistry
University of Cambridge
CB2 1EW, UK
+44-1223-763069

[Sam Adams]

On Thu, Nov 10, 2011 at 7:52 PM, Jorge Estrada <jorge....@unizar.es> wrote:

We are still studying how to add several data fields to Chempound.

We would like to see the following fields:

compchem:package
compchem:version
compchem:title
formula
n:Total energy/n:e_scf
n:wavefunction
n:charge

I'm going through the DepositNWChem code, and I see it generates the RDF that Chempound uses, but I still don't know how it generates the RDF, nor how to add our desired fields from the NWChem CML file.

How is that RDF data generated? How is that RDF data processed by the Chempound server and stored?

The NWChemLogImporter controls the RDF generation.  It calls the CmlComp2RdfConverter which generates RDF from CML conforming to the compchem convention, and then adds a few extra triples (e.g. the fact that it is an NWChem calculation).  The Chempound server records the all the files making up the entry in the RDF model (as an ORE Aggregation/ResourceMap), may do a little normalisation of the title and authors/contributors, and adds a created timestamp.  The RDF model is then stored in a triple store (currently Apache Jena, using the TDB data store).

 

How are the RDF data elements selected to appear on Chempound pages?

When a page gets viewed, Chempound checks the different 'handlers' it has installed and finds the one best suited to the entry's data type. The compchem-handler module is responsible for generating Chempound pages for compchem calculations.  It uses a freemarker template (comp.ftl) to generate the HTML.  The CompWriter class loads the template and provides the data model to freemarker.

 

I'm afraid this is a very brief summary of a fairly complex process... I don't have time to write it all up at the moment, so if you need to understand it you'll have to dig into the code a bit.  I hoping to find the time to refactor some of this over the next few weeks, as there's a fair bit of duplication in places, and to process could be simplified a bit.

Cheers,

Sam

[Peter Murray-Rust]

Many thanks Sam,
[I am now returned from Australia].

Just to re-emphasize how grateful we are to Sam for the Chempound software and advice. Sam is no longer working in the PMR group but is continuing to contribute to Chempound as he feels able.

Like all community collaborative projects we benefit from whatever can be put in whenever  by whomever. Sam wrote:

I'm afraid this is a very brief summary of a fairly complex process... I don't have time to write it all up at the moment, so if you need to understand it you'll have to dig into the code a bit.  I hoping to find the time to refactor some of this over the next few weeks, as there's a fair bit of duplication in places, and to process could be simplified a bit.

From my own experience Sam's code is very well thought out with a strong basis in modern software patterns. I think it will be valuable for us to document the current overall processes and modules on the basis that it will be refactored later. If anyone is intending to make major changes, then it will be a good idea to expose your ideas on the list so that sam can - hopefully - comment on whether it's a good idea.

The good news is that in my travels to PNNL and CSIRO  everyone is very excited about semantics in general and Chempound/Quixote in particular. And the package that we put together for the eRearch workshop shows that it works and can be distributed. I'll therefore be summarising the components and the distrib and asking people to try it out.

There are a number of things that people can help with. We need a bug tracker for the software. the bugs I know about are web-facing stuff (the stylesheet fails to work for firefox on my machine and the Java applet also fails - these may be specific to me or may be general). We need testing, customisation and evangelism (we can probably regard this as a V0.1). I am very appreciative also of Jorge's help - and commend his use of docbook for the documentation.

I will be going to CSIRO in Feb and working intensively there to get semantic physical science implemented there (copied to Nico Adams - do you subscribe to Quixote lists).

P.