Green Team CQs and Translator/Reasoner Intersections and Synergies

[karafecho]

Green Team has shifted our efforts on CQ2 from the Translator Jupyter notebooks to the PROTOCOP Reasoner. Non-technical team members have been able to successfully execute CQ2 (as an AOP) and generate scored and ranked answers. Those same team members are currently evaluating query output (i.e., answers), as well as the PROTOCOP UI/system. In parallel, technical team members are assessing query requirements and approaches for using the PROTOCOP Reasoner to answer (either in full or in part) other Green Team CQs. At this point, CQ3 appears to be suited for PROTOCOP, with modification of software and approach. CQ1 and CQ4 require a new data source, namely, patient data. Team members are exploring the use of clinical feature vectors (aka, clinical co-occurrence tables) as a new Translator Knowledge Source, one that hopefully will allow us to answer CQ1 and CQ4. CQ5 is dependent on successful execution of CQ4. CQ4 and CQ5 are being evaluated for potential execution via PROTOCOP.

[stevencox]

Demo'd this approach to NCATS.

[karafecho]

Received positive feedback from NCATS.

[karafecho]

Developed three new CQs related to Fanconi anemia and generated new ROBOKOP blackboards. Met with Maureen to discuss results. Received both positive and lukewarm feedback, depending on the specific CQ. Maureen's reaction was more positive for answers to really simple CQs (e.g., ethanol -> gene -> Fanconi anemia) than to more complex CQs (e.g., an ethanol to Fanconi anemia COP), even though the latter generated more results. Maureen is mostly interested in genetic relationships between socioenvironmental exposures and Fanconi anemia, however, so this may explain her lukewarm reaction to the more complex CQs.