search

trvrb / flux

Integrating influenza antigenic dynamics with molecular evolution
http://bedford.io/papers/bedford-flux/
12 stars 5 forks source link

BMDS priors #26

Closed trvrb closed 11 years ago

trvrb commented 11 years ago

What prior was used for the Baysian MDS? How does this choice of prior influence the resulting antigenic maps? How have the authors justified their choice of prior?

trvrb commented 11 years ago

We assume priors on several BMDS parameters: MDS variance φ2, drift rate μ, virus variance σx2 and serum variance σy2. We state in Materials and methods:

Top-level priors for 1/φ2, mu, 1/σx2, and 1/σy2 are assumed to follow diffuse Gamma( a, b ) distributions with a =0.001 and b = 0.001.

These are extremely diffuse priors and should have very little impact on the resulting antigenic maps. Running the analysis with improper uniform priors has almost no impact on the results. We've made note of this in the text.