Modeling noise with Gaussian Mixture Models (GMM)

Scales data in a Region object by fitting a GMM and using one of the model fit parameters to scale the data (see documentation for processing below).

Usage

model_norm_region(
  region,
  processing = c("divide", "asinh", "prob", "posterior"),
  rescale = NULL,
  to = "NormalizedData",
  from = "Data",
  ...
)

Arguments

region

A Region object, or concatenated data generated by .smapply.

processing

What should the function do with the model fits? Options are:

• "divide": Use the mean of the noise fit as a denominator for scaling values.

• "asinh": Use the mean of the noise fit as custom scale factors for an arcsinh transformation.

• "prob": Redefine values as the cumulative probability of coming from the signal fit.

• "posterior": Denoise values as Signal * PosteriorProbability(true distribution).

rescale

A function, optional, to additionally scale the GMM means.

to

Which data slot to assign the normalized data to.

from

Which data slot to pull data from.

...

Arguments passed on to model_norm

k

The number of mixture components

mu

The estimated means of the mixture models

sigma

The estimated standard deviations of the mixture models

lambda

The estimated proportions of the mixture models

fallback.fun

In case the mixture algorithm fails, what is the fallback function to define the noise distribution?

max.restarts

How many times the EM maximization should re-initiate if variances go to zero?

max.iter

How many iterations should the algorithm take to converge before giving up?

Value

An updated Region object with normalized data added to the slot specified in to. Metadata recording information about the normalization is also added to the featureInfo.

See also

Normalize()