CoGAPS {CoGAPS}R Documentation

CoGAPS Matrix Factorization Algorithm

Description

CoGAPS Matrix Factorization Algorithm

Usage

CoGAPS(D, S, nFactor = 7, nEquil = 1000, nSample = 1000,
  nOutputs = 1000, nSnapshots = 0, alphaA = 0.01, alphaP = 0.01,
  maxGibbmassA = 100, maxGibbmassP = 100, seed = -1, messages = TRUE,
  singleCellRNASeq = FALSE, whichMatrixFixed = "N",
  fixedPatterns = matrix(0), checkpointInterval = 0,
  checkpointFile = "gaps_checkpoint.out", ...)

Arguments

D

data matrix

S

uncertainty matrix (std devs for chi-squared of Log Likelihood)

nFactor

number of patterns (basis vectors, metagenes), which must be greater than or equal to the number of rows of FP

nEquil

number of iterations for burn-in

nSample

number of iterations for sampling

nOutputs

how often to print status into R by iterations

nSnapshots

the number of individual samples to capture

alphaA

sparsity parameter for A domain

alphaP

sparsity parameter for P domain

maxGibbmassA

limit truncated normal to max size

maxGibbmassP

limit truncated normal to max size

seed

a positive seed is used as-is, while any negative seed tells the algorithm to pick a seed based on the current time

messages

display progress messages

singleCellRNASeq

indicates if the data is single cell RNA-seq data

whichMatrixFixed

character to indicate whether A or P matric contains the fixed patterns

fixedPatterns

matrix of fixed values in either A or P matrix

checkpointInterval

time (in seconds) between creating a checkpoint

checkpointFile

name of the checkpoint file

...

keeps backwards compatibility with arguments from older versions

Details

calls the C++ MCMC code and performs Bayesian matrix factorization returning the two matrices that reconstruct the data matrix

Value

list with A and P matrix estimates

Examples

data(SimpSim)
result <- CoGAPS(SimpSim.D, SimpSim.S, nFactor=3, nOutputs=250)

[Package CoGAPS version 3.0.2 Index]