HCLSIG/LODD/Meetings/2010-11-10 Conference Call

From W3C Wiki
< HCLSIG‎ | LODD‎ | Meetings
Jump to: navigation, search

Conference Details

  • Date of Call: Wednesday November 10, 2010
  • Time of Call: 11:00am Eastern Time (ET), 16:00 British Time (BT), 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)
  • Duration: ~1h
  • Convener: Susie

Agenda

  • J Cheminformatics paper (Agenda items about the paper from Matthias below)
    • Finding consensus on style of the paper and target audience. I suggest that we focus on a high-level, not-too-technical view of the problem and our solution. Should appeal to all stakeholders.
    • Agreeing on overall paper structure (keeping in mind that we have a limit of only 1500 words for the main section).
    • Ask @Claus and @Susie if they could provide a paragraph about requirements from pharma perspective.
    • What is _part_ of LODD and what is only _related_ and connected to LODD, where do we draw the line? A bit difficult to decide for some data sets.
    • We should decide on a demo scenario. Neuropharmacology (including novel datasets, possibly ABA) would be a good candidate. Should we also use some of the queries presented by @Eric in his presentation?
    • Problems with integrating the Allen Brain Atlas (ABA) in its current version (the result of the GRDDL transform of ABA XML is only mildly useful for our purposes, since many key entities are not identified with URIs)
    • Interfaces we use for the demo: TripleMap, Sig.ma, maybe Exhibit (based on a simple script that digests SPARQL results). Any others?
    • @Chris, can you provide a short paragraph about TripleMap? Also, we might need your help formulating queries in TripleMap and providing screenshots once we have decided on a demo scenario.
    • How to deal with SPARQL queries and code -- I suggest placing most queries and code on a wiki page and reference it from the paper, instead of integrating them into the paper.
    • The paper should end with with a paragraph describing the next steps we plan with LODD. What are they? Could @Susie formulate this paragraph?
    • Discussion of other ideas by LODD participants.
  • Mapping experimental data
  • Data updates - Egon, Matthias, Anja, Oktie
  • AOB

Minutes

Attendees: Claus, Matthias, Egon, Elgar, Scott, Eric, Susie

Apologies: Bosse

    • Finding consensus on style of the paper and target audience. I suggest that we focus on a high-level, not-too-technical view of the problem and our solution. Should appeal to all stakeholders. -> People on the call agree.
    • Agreeing on overall paper structure (keeping in mind that we have a limit of only 1500 words for the main section). -> Going through the current Google-Doc. Nobodoy disagrees with the current overall structure.
    • Susie mentions PRISM. Proposal was put together by Phil Ashcroft (UCB). He uses LODD datasets. Maybe he could contribute to the paper.
    • Ask @Claus and @Susie if they could provide a paragraph about requirements from pharma perspective. -> Claus will start, Susie will edit later.
    • What is _part_ of LODD and what is only _related_ and connected to LODD, where do we draw the line? A bit difficult to decide for some data sets. -> Susie: core LODD datasets, perihperal LODD datasets (not involved, but communicated to us that they want to be listed). Maybe LODD could become an umbrella for other developments in the future (similar to OBO). It would make senste to describe in the paper that we want to become an umbrella group.

Scott: How should LODD "membership" be defined? Quality criteria? Matthias: Like OBO, but far more relaxed. Having a nice website and dataset overview is already a good start. Susie: open access for all, but we can offer best practices Matthias: could be built on CKAN

    • We should decide on a demo scenario. Neuropharmacology (including novel datasets, possibly ABA) would be a good candidate. Should we also use some of the queries presented by @Eric in his presentation?

Claus: would prefer neuropharmacology. Matthias: start with a compound, then see what data is available (e.g. drug targets -> brain regions -> disease associations) Scott: activity in different regions of the brain would be great.

    • Problems with integrating the Allen Brain Atlas (ABA) in its current version (the result of the GRDDL transform of ABA XML is only mildly useful for our purposes, since many key entities are not identified with URIs)

Susie: how about developing a use-case involving internal datasets. also including ADNI (locally done at J&J). Claus: Interesting. Susie: Image data can be added quantitatively, or you can download brain images. High-resolution MRI's. Also contains results of cognitive test. Different persons could have different versions of the disease. --> Susie and Claus write a paragraph about linking internal datasets.

All: discussion about disease area-- Alzheimer's vs. Parkinson's vs. Alzheimer's. Alzheimer's most interesting.


    • Interfaces we use for the demo: TripleMap, Sig.ma, maybe Exhibit (based on a simple script that digests SPARQL results). Any others? --> Claus offers that his group could help coding an Exhibit view.

Susie: Chris could also be interested in adding new datasets. Will ask him.


TODO: Susie to contact Phil. TODO: Claus provides use-case next week.