feature_selection: Feature selection

In jinghan1018/tensorICA: tensorICA

Description

Feature selection.

Usage

feature_selection(tica.o, component, topN = 0, CLkurt = 0)

Arguments

tica.o	The TICA result object, which can be obtained by function *DoTICA*.
component	The component ID.
topN	The number of top-ranked features to select from each inferred component.
CLkurt	The confidence level estimated by top-ranked features' kurtosis.

Value

pred.idx The list of predicted feature index.

k.p Visualizing the kurtosis of predicted feature.

References

Teschendorff AE, Han J, Paul D, Virta J, Nordhausen K. Tensorial Blind Source Separation for Improved Analysis of Multi-Omic Data. Genome Biology (2018) 19:76.

Examples

data(buccalbloodtensor);
Dim.l <- EstDim(buccalbloodtensor$data);
dim <- Dim.l$dim;
tica.o <- DoTICA(Data = buccalbloodtensor$data, dim = dim, method = "FOBI");
feature.n <- feature_selection(tica.o = tica.o, component = 12, topN = 62);
buccalbloodtensor$testDMCs[feature.n$pred.idx[[1]]];
buccalbloodtensor$testDMCs[feature.n$pred.idx[[2]]];
feature.k <- feature_selection(tica.o = tica.o, component = 12, CLkurt = 0.95);
buccalbloodtensor$testDMCs[feature.k$pred.idx[[1]]];
buccalbloodtensor$testDMCs[feature.k$pred.idx[[2]]];
feature.k$k.p[[1]];
feature.k$k.p[[2]];
 

jinghan1018/tensorICA documentation built on March 23, 2020, 5:26 a.m.