zinbFit: Fit a ZINB regression model

In drisso/zinbwave: Zero-Inflated Negative Binomial Model for RNA-Seq Data

Fit a ZINB regression model

Description

Given an object with the data, it fits a ZINB model.

Usage

zinbFit(Y, ...)

## S4 method for signature 'SummarizedExperiment'
zinbFit(
  Y,
  X,
  V,
  K,
  which_assay,
  commondispersion = TRUE,
  zeroinflation = TRUE,
  verbose = FALSE,
  nb.repeat.initialize = 2,
  maxiter.optimize = 25,
  stop.epsilon.optimize = 1e-04,
  BPPARAM = BiocParallel::bpparam(),
  ...
)

## S4 method for signature 'matrix'
zinbFit(
  Y,
  X,
  V,
  K,
  commondispersion = TRUE,
  zeroinflation = TRUE,
  verbose = FALSE,
  nb.repeat.initialize = 2,
  maxiter.optimize = 25,
  stop.epsilon.optimize = 1e-04,
  BPPARAM = BiocParallel::bpparam(),
  ...
)

## S4 method for signature 'dgCMatrix'
zinbFit(Y, ...)

Arguments

Y	The data (genes in rows, samples in columns).
...	Additional parameters to describe the model, see zinbModel.
X	The design matrix containing sample-level covariates, one sample per row. If missing, X will contain only an intercept. If Y is a SummarizedExperiment object, X can be a formula using the variables in the colData slot of Y.
V	The design matrix containing gene-level covariates, one gene per row. If missing, V will contain only an intercept. If Y is a SummarizedExperiment object, V can be a formula using the variables in the rowData slot of Y.
K	integer. Number of latent factors.
which_assay	numeric or character. Which assay of Y to use (only if Y is a SummarizedExperiment).
commondispersion	Whether or not a single dispersion for all features is estimated (default TRUE).
zeroinflation	Whether or not a ZINB model should be fitted. If FALSE, a negative binomial model is fitted instead.
verbose	Print helpful messages.
nb.repeat.initialize	Number of iterations for the initialization of beta_mu and gamma_mu.
maxiter.optimize	maximum number of iterations for the optimization step (default 25).
stop.epsilon.optimize	stopping criterion in the optimization step, when the relative gain in likelihood is below epsilon (default 0.0001).
BPPARAM	object of class bpparamClass that specifies the back-end to be used for computations. See bpparam for details.

Details

By default, i.e., if no arguments other than Y are passed, the model is fitted with an intercept for the regression across-samples and one intercept for the regression across genes, both for mu and for pi.

This means that by default the model is fitted with X_mu = X_pi = 1_n and V_mu = V_pi = 1_J. If the user explicitly passes the design matrices, this behavior is overwritten, i.e., the user needs to explicitly include the intercept in the design matrices.

If Y is a Summarized experiment, the function uses the assay named "counts", if any, or the first assay.

Currently, if Y is a sparseMatrix, this calls the zinbFit method on as.matrix(Y)

Value

An object of class ZinbModel that has been fitted by penalized maximum likelihood on the data.

Methods (by class)

• zinbFit(SummarizedExperiment): Y is a SummarizedExperiment.

• zinbFit(matrix): Y is a matrix of counts (genes in rows).

• zinbFit(dgCMatrix): Y is a sparse matrix of counts (genes in rows).

See Also

model.matrix.

Examples

se <- SummarizedExperiment(matrix(rpois(60, lambda=5), nrow=10, ncol=6),
                           colData = data.frame(bio = gl(2, 3)))

m <- zinbFit(se, X=model.matrix(~bio, data=colData(se)),
             BPPARAM=BiocParallel::SerialParam())
bio <- gl(2, 3)
m <- zinbFit(matrix(rpois(60, lambda=5), nrow=10, ncol=6),
             X=model.matrix(~bio), BPPARAM=BiocParallel::SerialParam())

drisso/zinbwave documentation built on March 18, 2024, 5:13 p.m.