MASH results on V8 data¶

Instead of setting $\hat{V} = cor(Z_{null})$ we set $\hat{V} = I$

Input data is generated by procedures documented here. Code chunk below runs the workflow.

!sos run analysis/20171002_MASH_V8.ipynb mash --vhat 0

library(lattice)
library(ggplot2)
library(colorRamps)
library(mashr)
library(repr)

Loading required package: ashr

res = readRDS('~/Documents/GTExV8/MASH/GTExV8.ciseQTL.4MASH.xtx.K5.P3.V0.mash_model.rds')
res$result = readRDS('~/Documents/GTExV8/MASH/GTExV8.ciseQTL.4MASH.xtx.K5.P3.V0.mash_posterior.rds')

MASH model fit¶

The log-likelihood of fit is:

get_loglik(res)

Here is a plot of weights learned.

options(repr.plot.width=12, repr.plot.height=4)
barplot(get_estimated_pi(res), las = 2, cex.names = 0.7)