Groups

[AnnG-Codes]

Could you also do the analysis without comparisons but looking at the entirety of the dataset? PhosR learns phosphorylation patterns across all samples, then uses group labels to interpret and contextualize those patterns

A) Global learning (data-driven)
kinase–substrate scoring
correlation structure
clustering

-> Uses all samples simultaneously

B) Group-based interpretation
differential phosphorylation
condition-specific kinase activity
biological conclusions

-> Uses your experimental design

[falconsmilie]

That's a fair way to think about it, with one caveat about current PhosPy scope.

In PhosPy today, explicit pairwise comparisons are optional rather than required. If no comparisons are supplied, the preprocessing path still runs and simply returns the corrected dataset without adding pairwise comparison columns. The downstream predMat workflow and the native kinase workflow then operate on the phosphosite-by-sample matrix itself, so analysis can already be performed across the full dataset rather than only through predefined contrasts.

That said, PhosPy 1.0.0 is still deliberately narrow in scope. More generally, this is broadly consistent with how PhosR is framed: kinase–substrate scoring is learned from the full phosphoproteomic matrix by combining motif matching with phosphorylation-profile matching across samples, while experimental design and group structure are then used to organise and interpret downstream biological results. (PMC; PhosR vignette)

PhosPy can already work on the entirety of the dataset without explicit comparisons, but the richer group-based interpretation layer is not yet as fully developed or exposed as a separate workflow in the current package.