search

saezlab / decoupleR

R package to infer biological activities from omics data using a collection of methods.
https://saezlab.github.io/decoupleR/
GNU General Public License v3.0
176 stars 23 forks source link

how to compute the pathway activity score for the pathway if I do not have the pathway genes weight #92

Closed FADHLyemen closed 8 months ago

FADHLyemen commented 10 months ago

Hi I am interested in computing PAS for the pathway activity score for the mtor from the Reactome database. I have 39 genes annotated under this pathway. unfortunately, decoubler needs weights weight for these genes which I do not know how to get them. any help? second, did you compare the performance of MLM with ssGSEA? Thank you

deeenes commented 10 months ago

Hi, The pathway activity inference is based on transcriptomics signatures, which are available for canonical pathways in the PROGENy database. For Reactome pathways no such signatures exist, hence it's not possible to infer their activities.

FADHLyemen commented 10 months ago

How did you infer the pathway genes weight? is it disease-specific? did you compare MLM with ssGSEA? Regards

deeenes commented 10 months ago

The weights have been derived by fitting models on transcriptomics data from dozens of targeted perturbation experiments, as described here. They are not disease specific. The other question @PauBadiaM will answer bc I'm not familiar with the topic.

PauBadiaM commented 9 months ago

Hi @FADHLyemen,

You can use ulm or mlm without weights, just add a column called mor in your net dataframe with 1s, like this:

net['mor'] <- 1

We did compare MLM and GSEA, and we found that any linear method (both ulm and mlm) outperforms classic enrichment analysis like ssGSEA (in this plot is fgsea) and GSVA (Fig 1C of the manuscript).