spatialNormalization {cellHTS2}R Documentation

Spatial normalization

Description

Correction of spatial plate effects of data stored in slot assayData of a cellHTS object by fitting a polynomial surface within each plate using local regression (loess or robust local fit). Uses a second degree polynomial (local quadratic fit). Only wells containing "sample" are considered to fit the models.

Usage

spatialNormalization(object, model="locfit", smoothPar=0.6, save.model=FALSE)

Arguments

object a cellHTS object that has already been configured.
model character indicated whether the polynomial surface should be fitted using robust "locfit" or "loess". The default is "locfit".
smoothPar numeric. The default is smoothPar=0.6. The parameter which controls the degree of smoothing (corresponds to 'span' argument of loess, or to the parameter 'nn' of lp of locfit).
save.model a logical value specifying whether the per-plate spatial effects should be stored in slot rowcol.effects of object. See details.

Details

For convenience, this function should be called indireclty from normalizePlates function.

The normalization is performed separately for each replicate and channel, in a per-plate fashion using the chosen local regression model. The polynomical surfaces are fitted using the wells containing "sample".

If save.model=TRUE, the models row and column offsets are stored in the slot rowcol.effects of object.

Value

An object of class cellHTS with normalized data stored in slot assayData. Furthermore, if save.model=TRUE, the row and column estimated offsets are stored in the slot rowcol.effects. This slot is a 3D array with the same dimension as Data(object).
Moreover, the processing status of the cellHTS object is updated in the slot state to object@state[["normalized"]]=TRUE.

Author(s)

Ligia Bras ligia@ebi.ac.uk

See Also

medpolish, loess, locfit, plotSpatialEffects, normalizePlates, summarizeChannels, plateEffects

Examples

    data(KcViabSmall)
    x <- KcViabSmall
    xs <- spatialNormalization(x, model="loess", save.model = TRUE)
    ## Calling spatialNormalization function from "normalizePlates":
    xopt <- normalizePlates(x, method="loess", varianceAdjust="none", save.model = TRUE)
    all(xs@rowcol.effects == xopt@rowcol.effects, na.rm=TRUE)

[Package cellHTS2 version 2.6.0 Index]