COMPASS (version 1.10.2)

SimpleCOMPASS: Fit the discrete COMPASS Model

Description

This function fits the COMPASS model from a user-provided set of stimulated / unstimulated matrices. See the NOTE for important details.

Usage

SimpleCOMPASS(n_s, n_u, meta, individual_id, sample_id, iterations = 10000, replications = 8, verbose = TRUE)

Arguments

n_s

The cell counts for stimulated cells.

n_u

The cell counts for unstimulated cells.

meta

A data.frame of metadata, describing the individuals in the experiment. Each row in meta should correspond to a row in data. There should be one row for each sample; i.e., one row for each element of n_s and n_u.

individual_id

The name of the vector in meta that denotes the individuals from which samples were drawn.

sample_id

The name of the vector in meta that denotes the samples. This vector should contain all of the names in the data input.

iterations

The number of iterations (per 'replication') to perform.

replications

The number of 'replications' to perform. In order to conserve memory, we only keep the model estimates from the last replication.

verbose

Boolean; if TRUE we output progress information.

Value

A list with class COMPASSResult with two components, the fit containing parameter estimates and parameter acceptance rates, and data containing the generated data used as input for the model.

Examples

## Not run: 
# set.seed(123)
# n <- 10 ## number of samples
# k <- 3 ## number of markers
# 
# ## generate some sample data
# sid_vec <- paste0("sid_", 1:n) ## sample ids; unique names used to denote samples
# iid_vec <- rep_len( paste0("iid_", 1:(n/2) ), n ) ## individual ids
# data <- replicate(n, {
#   nrow <- round(runif(1) * 1E4 + 1000)
#   ncol <- k
#   vals <- rexp( nrow * ncol, runif(1, 1E-5, 1E-3) )
#   vals[ vals < 2000 ] <- 0
#   output <- matrix(vals, nrow, ncol)
#   output <- output[ apply(output, 1, sum) > 0, ]
#   colnames(output) <- paste0("M", 1:k)
#   return(output)
# })
# meta <- data.frame(
#   sid=sid_vec,
#   iid=iid_vec,
#   trt=rep( c("Control", "Treatment"), each=(n/2) )
# )
# 
# ## generate counts for n_s, n_u
# n_s <- CellCounts( data[1:(n/2)], Combinations(k) )
# n_u <- CellCounts( data[(n/2+1):n], Combinations(k) )
# 
# ## A smaller number of iterations is used here for running speed;
# ## prefer using more iterations for a real fit
# SimpleCOMPASS(n_s, n_u, meta, "iid", "sid", iterations=100)
# ## End(Not run)