What is the shortest path to migrate Exomiser pipeline to uPheno2?

[matentzn]

The Exomiser pipeline is still using uPheno1 and an ancient mapping file p2p. That is why such complaints are now rightfully emerging:

[hhx464@frontend8 ~]$ zgrep HP_0002835 hp-hp-phenodigm-cache.txt.gz | grep HP_0002155

HP_0002835    HP_0002155    0.25   2.029360701409414    MP_0002133;

HP_0002155    HP_0002835    0.25   2.029360701409414    MP_0002133;
It seems to be finding a LCS of http://www.informatics.jax.org/vocab/mp_ontology/MP:0002133 which I agree makes no sense.
We really need to find a way to accelerate Nico and David’s work – so far we have had nothing to test ☹

[matentzn]

We need to remember that one of the biggest problems with the equivalence model is that it is very vulnerable to cross-ontology infection. If one ontology says X is a "abnormal respiratory system physiology" then anything mapped to it, and all the descendants from all branches with inherit that. Thats why I preferred the lattice model. A safer strategy than a complete equivalence model is to shadow the entire spectrum of phenotype in uPheno, map the sspos into uPheno 1:1 and only use uPheno for the semantic similarity analysis. That way you can be much more conservative with inheriting subclass axioms from phenotypes that do not have EQs attached with it. In fact, you can choose to not import any subclass axioms by default, and just allow undefined classes to float up (or use inherited partial EQ statements for some granular classification, like "inheres in some lung" -> abnormal anatomical entity.)