R: Parametrically Fit a Mean-Variance Curve

meanVarParaFit {MAnorm2}

R Documentation

Parametrically Fit a Mean-Variance Curve

Description

meanVarParaFit fits a mean-variance curve by applying a robust, gamma-family glm regression, taking advantage of the form: var = c1 + c2 / (2 ^ mean).

Usage

meanVarParaFit(
  x,
  y,
  weight,
  range.residual = c(1e-04, 15),
  max.iter = 50,
  init.coef = NULL,
  verbose = TRUE
)

Arguments

`x`	Two numeric vectors of (sample) means and sample variances, respectively.
`y`	Two numeric vectors of (sample) means and sample variances, respectively.
`weight`	An optional vector of weights to be used in the fitting procedure. It's typically used when sample variances in `y` are associated with different numbers of degrees of freedom.
`range.residual`	A length-two vector specifying the range of residuals of non-outliers.
`max.iter`	Maximum number of iteration times allowed during the fitting procedure.
`init.coef`	An optional length-two vector specifying the initial values of the coefficients.
`verbose`	Whether to print processing messages about iteratively fitting the mean-variance curve?

Details

meanVarParaFit iteratively detects outliers and fits a generalized linear model on non-outliers. The procedure converges as soon as the set of outlier points fixes.

See "References" for the theoretical foundation of the parametric form.

Value

A prediction function which accepts a vector of means and returns the predicted variances, with an attribute named "coefficients" attached.

References

Robinson, M.D. and G.K. Smyth, Small-sample estimation of negative binomial dispersion, with applications to SAGE data. Biostatistics, 2008. 9(2): p. 321-32.

Love, M.I., W. Huber, and S. Anders, Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol, 2014. 15(12): p. 550.

Tu, S., et al., MAnorm2 for quantitatively comparing groups of ChIP-seq samples. Genome Res, 2021. 31(1): p. 131-145.