Kinase Substrate Prediction

[AnnG-Codes]

As I understand, currently PhosR gets the kinase Substrate score mainly from PhosphoSitePlus (data: "PhosphoSitePlus") and some similar patterns within the read in data sets (?), however there are many more databases that could potentially be queried to get a better picture of what is happening e.g. PhosphoNetworks, PhosphoELM, GPS, etc. Maybe it would be nice to give the end user an option which data base to use

What Kinase Substrate Score means; ELI5:

Each cell has proteins. Proteins need to be regulated. Particularly this happens by sticking phosphate groups to a protein. Something (a protein) that can add a phosphate is called a kinase. Something (another protein, sometimes the same protein) that has a phosphate added to is called a Substrate. Kinase recognise their substrates through distinct motifs. These motifs are specific amino acid sequences/amino acid properties (basic or acidic or neutral amino acids can give rise to similar motifs). Different kinase groups look for different motifs. Sometimes motifs are shared between different kinases. The score predicts: How consistent is this phosphosite with being regulated by this kinase, given both prior knowledge and the experimental data?

[falconsmilie]

Thanks, this is a very good point.

Just to make the biology precise: in PhosR, the kinase–substrate score is not only a motif match. It combines motif evidence with phosphorylation-profile evidence from the experimental data. PhosR also makes use of prior kinase–substrate annotation resources such as PhosphoSitePlus and Phospho.ELM. (Kim et al., 2021; PhosR vignette) (PMC; PhosR vignette)

For PhosPy as it currently stands, the package does not yet offer a built-in option to choose between upstream resources such as PhosphoSitePlus, PhosphoELM, GPS, or PhosphoNetworks. In the predMat workflow it consumes a precomputed prediction matrix, and in the native workflow it works from the supplied substrate and motif inputs. (PhosPy analysis.py; PhosPy workflow.py) (GitHub analysis.py; GitHub workflow.py)

So I think your suggestion is a good one: PhosPy is currently flexible about the inputs it can work from, but it does not yet make the upstream knowledge source a first-class user option. Making that more explicit and configurable would be a sensible future enhancement. (GitHub analysis.py; GitHub workflow.py)