Drawing Networks, Interaction Networks, Social Networks
There is a cool presentation on google-tech-talks where Tamara Munzner introduce 15 algorithms describing how to draw a graph. That might be of interest for people working with "systems biology".
About those talks, the information is not displayed on google-video web site but it is now possible to get a RSS to keep track of the new talks at the following URL: http://video.google.com/videosearch?q=engEDU&so=1&output=rss (just add
&output=rss at the end).About networks, I've (re-)discovered a interesting source of data for protein-protein interactions. The database is called BioGrid(http://www.thebiogrid.org/index.php) and it was described in the database issue of NAR (BioGRID: a general repository for interaction datasets). I like this kind of resource because all interactions described in that database are manually curated from pubmed abstracts. This is much more rigourous than the large interaction maps screenings because, in the case of experiments using the yeast two hybrid system, for each distinct interaction, there should have some test to confirm the interaction. At the time I was using this assay, I used the method described by Bartel &. al (inverting the bait and the pray, co-immunoprecipitation, etc...) to confirm any interaction. But let's be fair, I have not read much bibliography about this subject since 1999...
The graphics from BioGrid can be exported in SVG...
....and the whole database can be downloaded in the PSI format.
PSI is a simple XML format describing
....interactors...
(...)
<proteinInteractor id="BIOGRID-109561">
<names>
<shortLabel>HGNC:5261</shortLabel>
</names>
<xref>
<primaryRef db="MIM" id="118190" secondary="" version=""/>
</xref>
<organism ncbiTaxId="9606">
<names>
<shortLabel>Human</shortLabel>
<fullName>Homo sapiens</fullName>
</names>
</organism>
</proteinInteractor>
(...)
... and interactions....
(...)
<interaction>
<names>
<shortLabel>1</shortLabel>
</names>
(...)
<participantList>
<proteinParticipant>
<proteinInteractorRef ref="BIOGRID-106718"/>
<role>bait</role>
</proteinParticipant>
<proteinParticipant>
<proteinInteractorRef ref="BIOGRID-106718"/>
<role>prey</role>
</proteinParticipant>
</participantList>
<interactionType>
<names>
<shortLabel>Invivo</shortLabel>
</names>
<xref>
<primaryRef db="" id="" secondary="" version=""/>
</xref>
</interactionType>
<xref>
<primaryRef db="pubmed" id="12081471" secondary="" version=""/>
</xref>
</interaction>
(...)
as far as I know,PSI is much more simplier than SBML or BIOPAX for describing simple protein-protein interactions (no biochemistry, no pKa, etc... however the format could be modernized by supporting a RDF/XML syntax). I also sent them a mail, asking how to sublit a new interaction but I still got no answer.
About social networks: David Wolber has introduced his social network in a google tech talk. There are a few good ideas about social networks: document and people are linked by a semantic property like in a FOAF document.
via google: Peoplicious is a collaborative research tool. Unlike systems such as del.icio.us, people are first-class data objects, along with documents. Users can create people, provide structured information about people (image, homepage, blog feed, delicious name, etc.), and create lists of people (people-tagging). Users can also bookmark documents and associate documents with people (personmarks). Any user can enter information about any person. Peoplicious is where del.icio.us meets RSS reader all within a modern day address book. There is a working prototype that can be accessed at: http://peoplicious.com/Technorati/lists. It is designed so that new sites can easily be created for particular domains (e.g., see http://peoplicious.com/USF/lists).
No comments:
Post a Comment