cellrPLR_biom: cell-rPLR biomarker identification

In walachja/cellrPLR: Cellwise Outlier Diagnostics Using Robust Pairwise Log-Ratios (cell-rPLR)

Description

Algorithm for cellwise outlier diagnostics using robust pairwise log-ratios - biomarker identification.

Usage

cellrPLR_biom(data, type = "biweight", g1, g2, mainGroup = "max",
              biomarker, permutation = FALSE, B = 1000, p.alpha = 0.95)
## S3 method for class 'biomarker'
print(x, ...)              

Arguments

data	dataset, either 'matrix' or 'data.frame'
type	type of weighting function, possible values are 'biweight', 'huber', 'hampel'
g1	vector with positions of samples from group 1
g2	vector with positions of samples from group 2
mainGroup	integer or character: group which is chosen as based. Possible values are: 1,'1' - group 1, 2, '2' - group2, 'all' - all samples, 'max' - the bigger group
biomarker	variable names of known biomarkers (not necessary)
permutation	logical value if permutation tests should be performed
B	number of permutation iterations, not used if permutation = FALSE
p.alpha	cut-off p-value for permutation tests, not used if permutation = FALSE

Value

List of Difference, Biomarker_results and Permutation_tests.

Difference	Difference value is returned. The bigger the value, more important variable is for discrimination. In range <0;2>.
Biomarker_results	Data frame with: Ordered_position, Biomarker_names and Variable_position.
Ordered_position	Variable importance ordered position.
Biomarker_names	If 'true' biomarkers were given as agrument biomarker, their names are returned.
Variable_position	If 'true' biomarkers were given as agrument biomarker, their ordered positions are returned.
Permutation_tests	Data frame with: p_value ,TRUE_FALSE,biomarkers,biomarkers_names.
p_value	If permutation = TRUE p_value for each variable is returned.
TRUE_FALSE	If permutation = TRUE infromation if variable is (TRUE) or is not (FALSE) identified as biomarker.
biomarkers_names	If permutation = TRUE names of identified biomarkers.
statistics	If permutation = TRUE and 'true' biomarkers were given: True positive (TP), True Positive Ratio (TPR), False Positive (FP) and False Positive Ratio (FPR) is given.

Author(s)

Jan Walach <[e-mail]>

References

'Cellwise outlier detection and biomarker identification in metabolomics based on pairwise log-ratios', Walach J., Filzmoser P., Kouril S., submitted

Examples

set.seed(452)
data <- gendata1_c()$X
colnames(data) <- LETTERS[1:9]
cellrPLR_biom(data, type = "biweight", g1 = 1:20 , g2 = 21:40, mainGroup = "all", biomarker=c('B','C'), permutation = TRUE, B = 100, p.alpha = 0.9)

walachja/cellrPLR documentation built on May 22, 2019, 2:47 p.m.