rank.pwer: rank.pwer : Calculate the power and the calibration of...

In jumentib/useFonc: Useful Functions

Description

rank.pwer : Calculate the power and the calibration of statiscal test (for simulation)

Usage

rank.pwer(pval, known.mediator = NULL, toplist = 20, decreasing = F,
  ral = 0.05)

Arguments

pval	pValues from a test
known.mediator	true result of the simulation
toplist	number of high pValues to use to calculate rank power
decreasing	default F. This function can be use with effects size and with decreasing = T
ral	default 0.05. Threshold for the calculation of the power and the recall

Details

AUC corresponds to the air under the curve (AUC) of a hits rank curve. This curve is obtained by looking for hits in a toplist of the 20 CpGs with the lowest pValues. F1 = 2*recall*power/(recall + power)

Value

cur : the rank power curve

auc : the AUC of the rank power curve

auc.max : the maximum AUC for a perfect model

auc.norm : auc/auc.max. Normalization of the AUC of the rank power curve

precision : statistical precision of the test : TP / (TP + FP)

recall : statistical recall of the test : TP / (TP + FN)

f1_score : F1 Score of the test : 2(Pre * Rec) / (Pre + Rec)

length.list : number of pValues lower than "ral" (Bonferroni correction).

TP : True positive

FP : False positive

FN : False negative

jumentib/useFonc documentation built on Nov. 18, 2019, 3:17 p.m.