fitModels: fitModels

In mcalgaro93/benchdamic: Benchmark of differential abundance methods on microbiome data

fitModels

Description

A wrapper function that fits the specified models for each taxon of the count data and computes the mean difference (MD) and zero probability difference (ZPD) between estimated and observed values.

Usage

fitModels(
  object,
  assay_name = "counts",
  models = c("NB", "ZINB", "DM", "ZIG", "HURDLE"),
  scale_HURDLE = c("default", "median"),
  verbose = TRUE
)

Arguments

object	a phyloseq object, a TreeSummarizedExperiment object, or a matrix of counts.
assay_name	the name of the assay to extract from the TreeSummarizedExperiment object (default assayName = "counts"). Not used if the input object is a phyloseq.
models	character vector which assumes the values NB, ZINB, DM, ZIG, and HURDLE.
scale_HURDLE	character vector, either median or default to choose between the median of the library size or one million to scale raw counts for the truncated gaussian hurdle model.
verbose	an optional logical value. If TRUE information on the steps of the algorithm is printed. Default verbose = TRUE.

Value

list of data.frame objects for each model. The first two columns contain the properly transformed observed values for mean and zero proportion, while the third and the fourth columns contain the estimated values for the mean and the zero rate respectively.

See Also

fitNB, fitZINB, fitDM, fitZIG, and fitHURDLE for the model estimations, prepareObserved for raw counts preparation, and meanDifferences for the Mean Difference (MD) and Zero Probability Difference (ZPD) computations.

Examples

# Generate some random counts
counts <- matrix(rnbinom(n = 600, size = 3, prob = 0.5), 
                 nrow = 100, ncol = 6)
# Estimate the counts assuming several distributions
GOF <- fitModels(
    object = counts, models = c(
        "NB", "ZINB",
        "DM", "ZIG", "HURDLE"
    ), scale_HURDLE = c("median", "default")
)

head(GOF)

mcalgaro93/benchdamic documentation built on Nov. 28, 2024, 2:16 p.m.