search

HUPO-PSI / mzIdentML

Repository for mzIdentML and the corresponding examples
23 stars 24 forks source link

cv mapping rules for Threshold? #131

Closed colin-combe closed 4 months ago

colin-combe commented 1 year ago

Hi, section 6.83 of the mzIdentML 1.2.0 spec (pg.81) defines Threshold as:

Depending on context: 1: The threshold(s) applied to determine that a result is significant. If multiple terms are used it is assumed that all conditions are satisfied by the passing results. 2: The threshold(s) applied to determine that a result is significant. If multiple terms are used it is assumed that all conditions are satisfied by the passing results.

...doesn't make a lot of sense, 1. and 2. are word for word the same.

Is one of the two contexts missing? Or is there only one definition and it does not depend on context?

(@andrewrobertjones @javizca )

Also (a more minor point), are there mistakes in the CV mapping rules for Threshold? E.g.:

MUST supply term MS:1001494 (no threshold) only once MUST supply term MS:1001448 (pep:FDR threshold) only once

Maybe I'm misinterpreting how to read these, but shouldn't that be MAY? Otherwise it's saying you must declare it has both no threshold and a threshold?

cheers, Col

colin-combe commented 1 year ago

i think some things need added to the cvParam mapping rules for Threshold possibly, any child of -

(e.g. MS:1001364 is given in example CV terms but i think wouldn't match any of the existing cv mapping rules?)

@lutzfischer?

colin-combe commented 1 year ago

actually, the two contexts of Threshold will be SpectrumIdentificationProtocol and ProteinDetectionProtocol... so there are two contexts and they are the same. My initial comment can be ignored.

still think there might be questions about the cv mapping rules for threshold

colin-combe commented 1 year ago

you could say that all the CV params that can go in Threshold should have an (indirect) "is a" relation to 'statistical threshold' (MS:1002482)?

colin-combe commented 1 year ago

you could say that all the CV params that can go in Threshold should have an (indirect) "is a" relation to 'statistical threshold' (MS:1002482)?

I'm wrong about that because not all thresholds are statistical thresholds.

The short, simple version of this issue is just that we would just like to add MS:1002664 (interaction score derived from cross-linking) to the CV mapping rules for Threshold.