Ntino › Comments

Eric Jain probably has some, and probably Pierre - Deepak Singh

Representing your data in RDF should be easy -- assuming it's already cleaned up and well structured, perhaps even stored in or exported to some XML representation. Don't even need any special tools for that. But if you are looking into setting up a complete ecosystem for editing, storing, searching, browsing etc, expect to invest a lot of time struggling with generic tools that want to do things just a bit different than you do... - Eric Jain

I've created a tool building a scientif network using pubmed/xml and foaf/rdf see http://plindenbaum.blogspot.com/2008... . It uses the Jena API. After a recent conversation with Alf Eaton I also (re) wrote a XSLT spreadsheet transforming pubmed to foaf http://code.google.com/p... - Pierre Lindenbaum

@Eric I agree. Representing our data in RDF is very easy. I'm just wondering what that gets us. We're evaluating the trade-offs for modeling our data different ways. Our immediate application is a web-based network visualization, but I'm also interested in the topic in a more general sense as well. - Sutee Dee

@Pierre Thanks for the example. Definitely will explore. Jena seems like a good choice. - Sutee Dee

RDF removes the need to invent your own conventions for representing a network/graph (e.g. on top of XML's tree model) + how to identify and reference resources across documents (XML only supports references within documents, and assumes documents). RDF may also allow you (or others!) to use generic tools e.g. for storing and searching the data. The drawback with RDF is that the standard XML syntax is verbose (but not much more than plain XML) and it's less straightforward to validate (at a schema level). - Eric Jain

simple scripts are out there for converting tab-delimited data (or from a relational db similarly) to RDF, by structuring triples using the table name as subject, column name as predicate, and cell data as object... can't spot my saved link for one, but Google "tab delimited to RDF"...also there's a bunch for XML to RDF -- what's the detail level of the network you want to represent? For specific RDF pathway modeling I'd suggest also biopax.org, there's good pointers there for conversion tools... - Ntino

@Sutee I know this thread is ancient on the FF timeline, but I'm curious what you and your group decided to do for this. Our lab is trying to answer the same question: how are we going to store and curate all the network data we commonly make use of? - Chris Lasher

@Chris We ended up storing the network data in a mysql database and outputting it as json from a web service. The graph data schema is basically edge_id, interactor_1, interactor_2, and optional columns for source and directed (boolean). We use entrez gene ids for the interactor columns. The web service response follows the format of the Google Visualization API... more... - Sutee Dee

Having read the above explanation, I want to like this again :-) - Neil Saunders

@Sutee Many thanks for taking the time to follow up. :-) Very helpful! @Neil I liked it on your behalf, since I missed this the first go 'round. - Chris Lasher