Using hierarchical models for High dimensional (protein expression) data

Modeling
, , ,
4

Hello!

As you mentioned, you could try the R2D2M2 prior. We have tested the R2D2M2 prior in high dimensional real life settings as you see in our paper here.

Here is a small minimal working example that shows how to set the prior and could help you:

library(brms
# Read the data
# I am using the popularity dataset, which I found in this brms tutorial
# https://www.rensvandeschoot.com/tutorials/brms-started/

popular2data <- read_sav(file = "https://github.com/MultiLevelAnalysis/Datasets-third-edition-Multilevel-book/blob/master/chapter%202/popularity/SPSS/popular2.sav?raw=true")

popular2data <- dplyr::select(popular2data, pupil, class, extrav, sex, texp, popular)

# Specifiy the R2D2 prior 
bprior <- prior(R2D2(mean_R2= 0.5, prec_R2=1 ,cons_D2 = 2, main = TRUE), class = sd) + 
  prior(R2D2(), class = b)

# Model with varying slopes

fit1 <- brm(popular ~ 1 + sex + extrav + texp + (1 + extrav | class),  
              data = popular2data, 
              prior = bprior,
              warmup = 1000,
              iter = 2000, 
              cores = 4, 
              chains = 4, 
              seed = 123) 

summary(fit1)
plot(fit1)
3 Likes