Bioc package: The Smart Transcriptomic Pipeline

knn_trn

R Documentation

knn_trn allows assessing the final DEGs through a machine learning step by using k-NN in a cross validation process.

Description

knn_trn allows assessing the final DEGs through a machine learning step by using k-NN in a cross validation process. This function applies a cross validation of n folds with representation of all classes in each fold. The 80% of the data are used for training and the 20% for test. An optimization of the k neighbours is done at the start of the process.

Usage

knn_trn(data, labels, vars_selected, numFold = 10, LOOCV = FALSE)

Arguments

`data`	The data parameter is an expression matrix or data.frame that contains the genes in the columns and the samples in the rows.
`labels`	A vector or factor that contains the labels for each of the samples in the data object.
`vars_selected`	The genes selected to classify by using them. It can be the final DEGs extracted with the function `DEGsExtraction` or a custom vector of genes. Furthermore, the ranking achieved by `featureSelection` function can be used as input of this parameter.
`numFold`	The number of folds to carry out in the cross validation process.
`LOOCV`	Logical parameter to choose between Loo-CV and KFold-CV.

Value

A list that contains seven objects. The confusion matrix for each fold, the accuracy, the sensitivity, the specificity and the F1-Scores for each fold and each genes, the best k found for the knn algorithm after tuning, and the predictions made.

Examples

dir <- system.file("extdata", package="KnowSeq")
load(paste(dir,"/expressionExample.RData",sep = ""))

knn_trn(t(DEGsMatrix)[,1:10],labels,rownames(DEGsMatrix)[1:10],3)

CasedUgr/KnowSeq documentation built on Aug. 16, 2022, 6:19 a.m.