Verification and Validating Testing - Real Data Use-Cases

[GregHydeDartmouth]

We capture verification and validation of various KP-specific use-cases in our test harness. The test harness will use both synthetic and real world data to validate we are making the correct inferences.

We plan on validating our system by asking our KP questions that have known answers documented in the literature (e.g., Trastuzumab yielding higher probability on survival/showing high positive relative contribution when querying our system over ERBB2 gene mutations.)

See https://github.com/di2ag/Connections-Hypothesis-Provider-Roadmap/milestone/9 for open issues.

[GregHydeDartmouth]

Use-case has changed slightly from initial comment. We are exploring if we can detect network motifs from gene expression data. Process involves building an initial adjacency matrix capturing pairwise associations between genes to used as the basis for BKB building. Validation involves comparing adjacency matrix with known gene-gene associations. Validation data is from Reactome interactions within Pathway Commons.