Make invalid HPO-IDs more prominent

[drseb]

Users might enter wrong HPO-IDs, because:

• use of a wrong version of HPO , i.e. use classes that do not yet exist in the HPO used by Exomiser
• use of term-IDs that belong to obsoleted terms in HPO, i.e. the HPO-ID does no longer map to a valid HPO-class
• entering wrong IDs, e.g. due to typos

Invalid HPO-IDs are left out by Exomiser (correct), but it seems hard to discover such cases for the user. Currently, only the log-file shows lines such as ... [main] ERROR de.charite.compbio.exomiser.core.prioritisers.PhenixPriority - invalid term-id given: HP:1002858 I could not find this information in the resulting HTML file for example.

This is IMHO an important information that users should get, because it might seriously affect the analysis results.

Any ideas?

[drseb]

The HTML even lists under Analysis settings:

...
pedPath: null
hpoIds:
- "HP:0000407"
- "HP:0009830"
- "HP:1002858"
modeOfInheritance: "UNINITIALIZED"
...

which gives the impression, that the given HPO-ID (HP:1002858) was actually used.

[julesjacobsen]

I agree they are an absolutely essential part of the system, although the Analysis settings shows what was submitted so it would be better to flag these up prominently in a seperate section. There is already an error section so this might be a good place to put them.

Given that there are other output formats too without such sections I'm not sure wether having a seperate error file or an error section header is a better idea.

Suggestions?

[visze]

My suggestion: I think if the user puts in an invalid HPO ID the analysis should not start. It makes no sense to exclude the HPO-ID because we do not know if it is the main phenotypic feature...

[julesjacobsen]

@visze I like it, plus it makes it much easier to code. InvalidHpoException extends RuntimeException or maybe add an HPO validation step which will do things a slightly less catastrophic manner.

[drseb]

Is it possible that we currently have multiple versions of HPO in Exomiser?

[drseb]

Just mentioning this, because then the validation is dependent on the selected prioritiser.

[julesjacobsen]

Yes - I think there are probably two versions. There is the one which is shipped in the data/phenix directory and another version used in the database for the HP-HP/MP/ZP associations. This is far from ideal.

If at all possible I'd like it so that the ontologiues and associations/phenodigm scores can be updated more easily, even by users directly.