Re: Towards a cyberinfrastructure for the biological sciences: progress, visions and challenges

----- "Matthias Samwald" <samwald@gmx.at> wrote:

> From: "Matthias Samwald" <samwald@gmx.at>
> To: "public-semweb-lifesci" <public-semweb-lifesci@w3.org>
> Sent: Wednesday, 20 August, 2008 6:53:09 PM GMT +10:00 Brisbane
> Subject: Re: Towards a cyberinfrastructure for the biological sciences: progress,  visions and challenges
>
> Kei Cheung wrote:
> > Also, it's interesting to see scientific workflows can be published
> via 
> > Wiki (e.g., myExperiment).
> 
> But as far as I know, myExperiment does not allow editing the actual 
> workflows online, you can only upload and visualize workflow files
> that have 
> been created on the client-side. I guess that still poses a
> significant 
> hindrance to realizing the 'anyone can edit' philosophy of classic
> wikis. In 
> this regard, fully server-sided systems such as the well known Yahoo
> Pipes 
> or the quickly maturing Semantic Web Pipes [1] might be the way to
> go.

I hadn't looked at DERI pipes for a while, but it has improved a lot since last time. It would be nice to get to the stage where the control flow can be supported in the same way as Yahoo Pipes. I wonder if Yahoo Pipes will eventually support SPARQL and RDF in the same way they support Atom etc now. Does anyone know whether can DERI pipes be output in a way that Yahoo Pipes (or another mashup tool) can use for input?

It would be nice for MyExperiment to be able to host generic mashups, ie, include Yahoo, Google Mashups, Popfly, DERI pipes and any others that scientists are using for their work. I don't think that it is workable for MyExperiment to host a particular active editor for workflows or mashups, but a nice repository of links and descriptions would be nice, possibly with appropriate static diagrams similar too the SCUFL generated diagrams. You can still copy and paste, and reupload your new workflows even if you can't edit them online. I think the fact that wiki's don't have any external dependencies, as opposed to scufl or mashups which have other sites as direct dependencies, makes them a lot easier to support so it might not be fair to refer to the basic text wiki as a framework that scientists will actually end up getting for their computational work, although for basic text file outputs from computation it might be successful.

If someone is able to create a SCUFL workflow, I think they can be assumed to also be able to create a sufficiently functional mashup as long as the development tools are similar. It is interesting how much of a typical scientific SCUFL workflow seems to consist of data grabbing and data binding, which can much more easily be done using RDF, IMO.

> Regarding the article, it will probably seem a bit puzzling to many
> people 
> on this mailing list that Lincoln Stein writes
> 
> "To my knowledge, there is currently only one project that aims to
> bring the 
> pure semantic web to biomedical research. That project is the Simple 
> Semantic Web Architecture and Protocol (SSWAP30)"

Wow, I wonder when he wrote this and whether he has read [3]. I can't see Bio2RDF failing his criteria for bringing the semantic web to biomedical research. Not sure what his criteria for pure are but I have always though it fits what Bio2RDF, and YeastHub (to pick one that I can think of right away) are doing. The set of projects that he outlines are mostly well known anyway, except for the SSWAP30 thing which might be worth looking into further as I hadn't heard of it before. Maybe getting into the details of the other larger non-semantic, well funded databases took his attention away from semweb, although Uniprot has a reasonable semantic presence so that shouldn't have passed his radar. Does anyone know whether Stein subscribes to this list?

> It's nice that Nature allows the community to add descriptions about
> our 
> various projects on the wiki page associated with the article [2]; 
> unfortunately, though, most readers of the original article will
> probably 
> not have a look at that wiki.

I think the idea of a comments section, if not a wiki page, for each article on a publishers site, or on PubMed maybe, would definitely provide a novel area of discussion. One type of site I can think of that does this well is the PHP documentation pages, which include examples and discussion right below the officially published API.

> [1] http://pipes.deri.org/
> [2]
> http://nrgwiki.nature.com/cyberinfrastructureforbiology/show/HomePage
> 
> Cheers,
> Matthias Samwald
> DERI Galway, Ireland // Semantic Web Company, Vienna


Cheers,

Peter

[3] http://www.ncbi.nlm.nih.gov/pubmed/18358788

Received on Wednesday, 20 August 2008 10:07:55 UTC