Targeted gene set models

[envest]

We have a gene set derived from the Nanostring nCounter PanCancer IO 360™ Panel. It's possible to download the gene list of this panel from here.

The question here is: can we use this targeted gene panel to build a model? How well does a random gene set of the same size do (negative control)?

[envest]

Experimental design:

• Run models with targeted Nanostring gene set using same train/test allocations as baseline models
• Run models with random gene sets (same size as Nanostring gene set) using same train/test allocation as baseline models
  • Repeat this 10x
• Visualization:
  • Two panels, one for targeted gene set, one for random gene sets
  • For each train/test allocation, define delta-kappa = (Targeted or Random kappa) - (Full model kappa)

Is delta kappa close to zero for targeted gene set? Is delta kappa greater (more negative(?)) for random gene sets?

@jaclyn-taroni 👋