FUNLDA (version 1.1)

FitLDAModel: fit an LDA model

Description

fit an LDA model

Usage

FitLDAModel(training.data, nclust = 9, kde.nbins = 1000, iters = 250,
  inner.iters = 200)

Arguments

training.data

A data frame with character-valued columns rs and cat and numeric-valued columns with annotations. Each row is data for one SNP in one tissue. rs is an ID for the SNP, which need not be unique, and cat is an ID for each tissue.

nclust

Integer specifying the number of clusters to estimate

kde.nbins

Integer specifying how many bins to use for binning each annotation

iters

Integer specifying number of outer iterations

inner.iters

Integer specifying number of inner iterations

Value

A fitted LDA model, i.e., a list (apart from elements used internally) with elements

p.labeled

a data frame with one row per training variant, with the posterior probability for each variant to be in each cluster in columns CLUSTER1,... and also with columns cat and rs

a.labeled

a data frame with one row per tissue with membership vectors for each tissue with columns cat and CLUSTER1,...

Examples

# NOT RUN {
  data(training)
  summary.c <- FitLDAModel(training.data=training, nclust=3,
                         kde.nbins=100, iters=50, inner.iters=50)
# }