HCLSIG/LODD/Meetings/2010-09-15 Conference Call
Conference Details
- Date of Call: Wednesday September 15, 2010
- Time of Call: 11:00am Eastern Daylight Time (EDT), 16:00 British Summer Time (BST), 17:00 Central European Time (CET)
- Dial-In #: +1.617.761.6200 (Cambridge, MA)
- Dial-In #: +33.4.26.46.79.03 (Paris, France)
- Dial-In #: +44.203.318.0479 (London, UK)
- Participant Access Code: 4257 ("HCLS").
- IRC Channel: irc.w3.org port 6665 channel #HCLS (see W3C IRC page for details, or see Web IRC)
- LiveMeeting URL: https://www.livemeeting.com/cc/jnjamericas/join?id=G7PP87&role=present&pw=5K%5BDH%7Bp
- Duration: ~1h
- Convener: Susie
Agenda
- TripleMap demo - Chris
- Mapping experimental data - Susie
- Data updates - Egon, Matthias, Anja, Oktie, Janos
- AOB
Minutes
Date: September 15, 2010
Attendees: Bosse, Julia, Susie, Chris Bouton, Claus, Egon, Jun, Scott
Chris presenting TripleMap Demo (Entagen, LLC)
* Entagen is a Semantic Tech company, building a system called TripleMap and using LODD data for a demo. * Aim to provide “dynamic knowledge maps” * 16 software engineers and informaticians * Large pharma clients * Web app for navigating triple stores via standard web browser * Rich graphics * SPARQL queries * RDF triple store * D2R to map to relational * SharePoint interface to Team Sites and ERooms * Less “graph-based” view of data and more of a “mind map” view * Users can build dynamic maps (as new info is generated, it’s added dynamically) for ongoing projects * Constant back-end literature scanning * Entities (proteins, compounds, essays, pathways) have properties and include hyperlinks to other web-based systems * To avoid “visual hairballs”, data can be presented in a sortable table or as “visual collections” * Services-based with API, to connect to tools such as rendering, analytics, modeling, virtual screening * Inferencing and network connectivity analysis * Maps can be shared based on group authentication and searched using network analytics * Can print out reports and summaries in PDF or Powerpoint * Question: how to deal with refutation in Pubmed articles? Can it be represented? Answer: users have access to original data sources. A link’s validity is based on the number of sources. * Question: is there chemistry development? Answer: yes * Question: did you convert any of the data? Answer: used LODD sets * Question: how to represent inferred connections and distinguish them from other types? Answer: no, only provided triples are currently used via SPARQL endpoint of the original source. * Demo: * Interface is based on the “master list” of entity types * Once entity is selected, all the properties are visible * SPARQL call then runs to identify all the relationships * Related entities can be dragged to workspace on demand (either as table or a visual collection – all compounds are rendered on the fly) * The system avoids duplicates on any given map * Information can come from different sources behind the scenes and be displayed together * The interface can be “touchable” * Question: how does the system decide if two entities are the same? * Answer: for a any given data source, the users have an interface to map entities and create a “master index” * For any two entities, the system traverses the tree on the fly to identify all potential connections * Maps can be integrated (links between canonical maps for two different diseases, for example) * Working on visualizing images and plate-based data * Susie: it would be good for LODD to use this software to demonstrate value * Chris: there is interest in adding patient data for personalized medicine analysis * Susie: to get a demo acct, send email to Susie and she will pass it on to Chris to get a “trial code” for testing * Chris: keep in mind that the system is still being heavily developed, so it might be down occasionally.