HCLSIG/PharmaOntology/Meetings/2010-02-02 Conference Call
Conference Details
- Date of Call: Tuesday, February 2 2010
- Time of Call: 10-11am EST
- Duration: 1h
- Convener: Elgar
Agenda
- next steps for TMO development
- tighter integration of TMO with your triplestore
- visualizations
- drug code mappings
Minutes
Attendees: Colin Batchelor (C), Michel Dumontier (M), Joanne Luciano (J), Elgar Pichler (E)
Apologies: Susie Stephens
- E:
- intro referring to agenda;
- interested in catching up on triple store work, coordinating next steps &
- efforts, preparing presentation of TMO work for CSHALS2010;
- M:
- problem we face: no ontology & data connect;
- reconcile what we actually have [in triplestore];
- interesting patient record data;
- unfortunately for other data no actual linking;
- queries so far based mostly on regexp matching;
- added NDC (National Drug Code) codes to triplestore;
- E:
- can try RxNorm matching to NDC based on mapping in UMLS;
- J:
- at Partners/Dana Farber have colleague [name?] who has experience with ontologies/mapping exercises - could contact him for help
- E:
- can spend some time on UMLS based mappings;
- M:
- if we have matches, then we can move away from (queries based on) matching on labels;
- inclusion/exclusion criteria: would be interesting to look at formalized criteria if they exist; otherwise create ontology to define propositions for in/exclusion criteria; see, e.g., emailed ontology;
- problem for ClinicalTrials.gov;
- can we extract propositions from other systems and align/normalize those;
- this is a difficult/hard problem because of different encoding, criteria used in different systems
- C:
- agreed, combinatorial problem, similar case for [?]
- M:
- prefer finding people who have done some of this;
- verify with those experts if on the right track;
- J:
- have done similar work in [what?]
- M:
- difficulties in these exercises have been highlighted by others; see paper by Patau (@IBM Watson) [add link?];
- SNOMED-BFO notation differences exist - how to align the two?
- C:
- SNOMED people are working on this;
- could talk to Kent Spackman at SNOMED;
- M:
- would be good to start a dialogue with SNOMED folks;
- C:
- see paper by Stefan Schultz [add link?] on topic;
- E:
- would be interesting to talk about mapping problems at CSHALS, and how efficient various approaches are in getting mappings;
- M:
- access to anonymized patient data would be nice;
- NW eHealth, partnering with UK eLab [correct names?, add links to what they are doing?]
- J:
- will contact MITRE contacts on anonymized data, see if some are available;
- M:
- will meet with CTO [of ?] to discuss which access to patient data is available and under what conditions these data would be available;
- general comment: paper was focussed on patient; also have opportunity to look at basic reasearch questions;
- other aspect is visualization: no good visualization interfaces exist for linked data; might have to build customized webapp for that;
- E:
- try to stay away from developing webapps unless we have resources for that;
- would prefer getting (potentially large size) dumps of multidim data sets based on certain queries from triplestore and then do multifaceted visual browsing/querying/display of data with standard visualization tools;
- description of visualization tools;
- problem: what happens if query result tables are very large? need further discussion on that;
- M:
- should also get input on visualization from speci-alists;
- E:
- agreed; not all visualizations make sense for everyone;
- probably will require iterative process to find good visualizations;
- how about tighter TMO-triplestore integration?
- M:
- URL to instantiated classes in triple store exists:
- which TMO classes are in triple store? after mapping them one would be able to do queries based on TMO;
- not currently working on that;
- M&J:
- discussion on content at
- http://tinyurl.com/yks8ms6 (graphs in triple store)
- http://tinyurl.com/yf464yz (instantiated classes in triple store)
- [experiencing technical problems with TC ...]
- E:
- thanks for first overview;
- will need a few days to go over information, recently communicated links and docs;
- will try to set up follow-up meeting for beginning of next week;