Tasks for new methods of phenotype matching and scoring are defined in our team operations "Phenotype matching algorithm" project. This work to improve the underlying PhenoDigm and IMPC databases should be done in conjunction with Exomiser dev. The main tasks are:

• get OWLSim3-bayesian running and test it improves performance
• replace the PhenoDigm scores with a more informative prob value
• incorporate use of negative phenotypes and phenotype annotation freqs
• replace lots of the pre-calculation of scores and hiPhive prioritiser code with direct use of the OWLSim3 library
• Peter is also keen to explore likelihood ration tests for known disease genes at least. Will give an odds-ration and post-test probability that clinical users are used to working with. Can generate ratio for a phenotype term based on prob(given disease) from Orphanet freqs and p(not given disease) from IC typs measures. Similar approach for variant scoring.