smfishHmrf.generate.centroid.it {smfishHmrf}

Generate cluster centroids, where input is a file.

Description

This function generates cluster centroids from applying kmeans. It accepts an expression matrix file as input.

Usage

smfishHmrf.generate.centroid.it(
  expr_file,
  par_k,
  par_seed = -1,
  nstart,
  name = "test",
  output_dir = "."
)

Arguments

expr_file

expression matrix file. The expression file should be a space-separated file. The rows are genes. The columns are cells. There is no header row. The first column is a gene index (ranges from 1 to the number of genes). Note the first column is not gene name.

par_k

number of clusters

par_seed

random generator seed (-1 if no fixing). Change the par_seed to vary the initialization.

nstart

number of starts (kmeans). It is recommended to set nstart to at least 100 (preferrably 1000).

name

name of this run

output_dir

output directory; where to store the kmeans results

Details

Note that after running kmeans step, the function also automatically saves the kmeans results to the output_dir directory. The results will be found in two files:

1. {output_dir}/k_{par_k}/f{name}.gene.ALL.centroid.txt

2. {output_dir}/k_{par_k}/f{name}.gene.ALL.kmeans.txt

where {} refers to the value of parameters.

Value

A kmeans object which is a list with centers and cluster fields

Examples

mem_file = system.file("extdata", "ftest.expression.txt", package="smfishHmrf")
kk = smfishHmrf.generate.centroid.it(mem_file, par_k=9, par_seed=100, 
    nstart=100, name="test", output_dir=tempdir())