Skip to content

Bayesian combinatorial multi-study factor analysis

Notifications You must be signed in to change notification settings

igrabski/tetris

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 

Repository files navigation

Tetris

Tetris performs Bayesian combinatorial multi-study factor analysis (see our preprint here) in R.

Usage

To run Tetris, you should first prepare your data as a list of matrices, where each matrix corresponds to data from a different study/group and is in samples by features format. You can then run the sampler with the following command:

# X is a list of matrices
# alpha, beta are hyperparameters 
run <- tetris(X,alpha=5,beta=1)

To recover point estimates of the factor indicator matrix and loadings matrix, you can run:

# alpha_IBP is the hyperparameter; S is the number of studies
A <- choose.A(run,alpha_IBP=5,S=3)
run_fixed <- tetris(X,alpha=5,beta=1,fixed=TRUE,A_fixed=A)
Lambda <- getLambda(run_fixed,A)

Tetris also has a clustering extension, which simultaneously estimates group structure alongside factor analysis parameters. To run this extension, you should prepare your data as a single matrix in samples by features format, and run:

# X is a matrix
# S is the number of clusters you want to find
# alpha, beta are hyperparameters
run_clustering <- tetris_clustering(X,S=3,alpha=5,beta=1)

About

Bayesian combinatorial multi-study factor analysis

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages