get.th.inv.set: Compute thresholds that define core sets per replicate

In rxmenezes/rscreenorm: Normalization of Multiple Genetic Screens Data

Description

This is an internal function to compute thresholds to be used later on for constructing core sets. Two methods are available: using the distance between negative and positive controls to determine the core set, and including in the set a proportion of the lethality scores. If no well annotation is available indicating which wells correspond to assay controls, then core sets are computed using a proportion of all available features. To be used internally, it is called by get.inv.set.

Usage

get.th.inv.set(data_rscreen, my_gamma = NULL, var_type = c("mad", "IQR",
  "sd"), prop = 0.95, set_shape = c("left", "centre"))

Arguments

data_rscreen	an object of class 'rscreen.object', either yielded by read.screen.data, or by combine.screens. Typically the 'data_only' slot of this object contains lethality scores computed by get.leth.scores, which already reflect a value relative to negative and positive controls, per replicate.
my_gamma	scalar or numeric vector, with value(s) between 0 and a finite number. Its default value is NULL. If given, it will be used to construct the core sets by using the relative distance between negative and positive controls' distributions. If a numeric vector is given, it must contain as many entries as columns in the data_only slot of data_rscreen. A value of 1 means that the core set includes all lethality scores up to the ratio between the MAD of negative controls, and the sum of the MADs of negative and positive controls, assuming the robust variability measure MAD is used.
var_type	string indicating the statistic used as variability measure. Possible values are 'mad' when the median absolute deviation (MAD) is used (the default), 'IQR' when the inter-quartile range is used, and "sd" when the standard deviation is used.
prop	scalar between 0 and 1, corresponding to the desired proportion of lethality scores to be included in the core set, per replicate. The default value is 0.95. If my_gamma is numeric, any value for this argument is ignored.
set_shape	string indicating the shape of the core set. It accepts "left" (the default), when all scores to the left of the set threshold are included, per replicate, and "centre", when scores included in the core set only exclude the largest and smallest ones, at equal frequencies.

Value

A list with the following elements: the first with the thresholds defining the core sets (one value or pair per sample, so either a vector or a matrix),

See Also

get.inv.set the wrapper that calls this functions and others to yield the resulting core sets, get.rscreenorm to normalize data from multiple screens using core sets and get.leth.scores to compute scores that make observations in different screens comparable.

rxmenezes/rscreenorm documentation built on May 15, 2019, 1:19 p.m.