search

greenelab / mpmp

Multimodal Pan-cancer Mutation Prediction
BSD 3-Clause "New" or "Revised" License
7 stars 6 forks source link

Explore gene dosage effects #74

Closed jjc2718 closed 2 years ago

jjc2718 commented 2 years ago

For our submitted paper, one of the reviewers asked how important the expression levels of the mutated gene (i.e. "gene dosage") are in our predictions: see greenelab/mpmp-manuscript#45.

The answer, for a limited set of genes, seems to be "generally not that important" (green and orange boxes similar and green and purple boxes different):

image

See the notebook 02_classify_mutations/dosage_effects/plot_dosage_effects.ipynb for additional details.

jjc2718 commented 2 years ago

Looks good! I wonder why gene dosage is relevant in some genes and not in others. Maybe they're more isolated in the coexpression network, so other genes' values aren't able to compensate for them as easily?

Yeah, it could be something like that. EGFR and ERBB2 are both cell surface receptors that are relatively "upstream" in their cancer-associated pathways, so maybe they have more unique information about cell signaling/tumor state etc. while more "downstream" genes (the rest of them AFAIK) are more redundant with the rest of their cancer pathway. I dunno, I'll have to think through it more.