compareROC2: get ROC Curve of one or more variates for a binary status

In shijianasdf/BasicBioinformaticsAnalysisFromZhongShan: Launch Usual Clinical tool for Kind Yield

Description

compareROC2 helps get ROC curves of one or more variates for a binary status.It supports the merge of lots of variates via glm function.

Usage

compareROC2(data, markers, status, merge.markers = NULL,
  roc.type = "ggplot", title = "ROC Curves", color = NULL,
  half.border = F, reference = T, legend.position = c(0.8, 0.2),
  show.auc = T, auc.digits = 2, width = 10, height = 10,
  names = "love")

Arguments

data	a data frame
markers	target markers
status	the dependent variable in ROC.like "N.status"
merge.markers	A list of markers you want to merge.Default is NULL.The merge stragegy is based on Generalized Linear Models(glm)
roc.type	one of "ggplot" and "pROC"."ggplot" is recommanded and the default setting
title	the plot title
color	Default is NULL.You can set other colors like "#8DD3C7"
half.border	whether to show half border style
reference	whether to show a reference line in ROC plot
legend.position	the position of legend in ggplot
show.auc	whether to show related auc in legend labels
auc.digits	the digits of auc in legend labels
width, height	the size of saved PDF plot
names	part of saved PDF plot names

Author(s)

Weibin Huang<[e-mail]>

Examples

## This is a simulative process and NOT RUN
train.LM.ROC <- compareROC2(
 data = design.x,
 markers = mgenes.s,
 status ="lymphatic.metastasis",
 merge.markers=list(merge = mgenes.s),
 roc.type = "ggplot",
 title="ROC Curves of Lymphatic Metastasis",
 color = mycolor[c(1,20,3,5,6,7,4)],
 legend.position=c(0.8,0.2),
 show.auc = T,auc.digits = 2,
 width = 10,height = 10,
 names =paste0("seed",seed,"_train.mgenes")
)
## Quick Start
data(fat)
markers=c("BMI","Waist","WHR","belly fat thickness") #marker #'colnames
status =  "outcomes" #survival status
merge.markers = list(c("BMI","Waist"))# the markers list that you want to merge as a co-prognostic factor
return.plot = T # if T，return plot;if F，return data
roc.type = c("ggplot","pROC")[1] # ggplot style is recommanded
title="this is a title" # plot title
color = brewer.pal(12, "Set3")[1:5] #curve colors.The lenght of colors must >= the number of curves.
output.name = "ROC of something" # part of PDF name
compareROC2(data=fat,
           markers,
           status,
           merge.markers=NULL,
           return.plot=T,
           roc.type,
           title,
           color,
           output.name="test1")

shijianasdf/BasicBioinformaticsAnalysisFromZhongShan documentation built on Jan. 3, 2020, 10:08 p.m.