R语言绘制ROC曲线
欢迎来到医科研,这里是白介素2的读书笔记,跟我一起聊临床与科研的故事, 生物医学数据挖掘,R语言,TCGA、GEO数据挖掘。
介绍其它用于ROC绘制的R包
library(tidyverse) # for data manipulation
library(pkgsearch) # for searching packages
找与ROC相关的包
该包会提供一系列关于感兴趣主题的R包,包括他们的评分,作者,连接等等
ps函数等价于pkg_search
size:定义返回结果数量
format="short"返回格式
Sys.setlocale('LC_ALL','C')
rocPkg <- pkg_search(query="ROC",size=200)
head(rocPkg)
class(rocPkg)
[1] "C"
- "ROC" ------------------------------------- 74 packages in 0.01 seconds -
# package version
1 100 pROC 1.15.0
2 44 caTools 1.17.1.2
3 18 survivalROC 1.0.3
4 18 PRROC 1.3.1
5 15 plotROC 2.2.1
6 14 precrec 0.10.1
by @
Xavier Robin 2M
ORPHANED 4M
Paramita Saha-Chaudhuri<U+000a><paramita.sahachaudhuri.work@gmail.com> 7y
Jan Grau 1y
Michael C. Sachs 1y
Takaya Saito 3M
[1] "pkg_search_result" "tbl_df" "tbl"
[4] "data.frame"
ROCR包
performance函数计算tpr,fpr
library(ROCR)
data(ROCR.simple)
df <- data.frame(ROCR.simple)
head(df)
## predictions labels
## 1 0.6125478 1
## 2 0.3642710 1
## 3 0.4321361 0
## 4 0.1402911 0
## 5 0.3848959 0
## 6 0.2444155 1
pred <- prediction(df$predictions, df$labels)
perf <- performance(pred,"tpr","fpr")
perf
plot(perf,colorize=TRUE)
plotROC包-ggplot绘制ROC曲线
ROC曲线用于评估连续测量的精度,以预测二进制结果。在医学上,ROC曲线用于评价放射学和一般诊断的诊断试验有着悠久的历史。ROC曲线在信号检测理论中也有很长的应用历史。
require(plotROC)
提供网页版操作,为了代码的连贯性,这里不介绍网页版,不可能我们分析到一般导出数据,拿到网页版去操作。
基本用法
set.seed(2529)
D.ex <- rbinom(200, size = 1, prob = .5)
M1 <- rnorm(200, mean = D.ex, sd = .65)
M2 <- rnorm(200, mean = D.ex, sd = 1.5)
test <- data.frame(D = D.ex, D.str = c("Healthy", "Ill")[D.ex + 1],
M1 = M1, M2 = M2, stringsAsFactors = FALSE)
head(test)
## D D.str M1 M2
## 1 1 Ill 1.48117155 -2.50636605
## 2 1 Ill 0.61994478 1.46861033
## 3 0 Healthy 0.57613345 0.07532573
## 4 1 Ill 0.85433197 2.41997703
## 5 0 Healthy 0.05258342 0.01863718
## 6 1 Ill 0.66703989 0.24732453
geom_roc绘图
d为编码1/0, m为用于预测的值marker
注意需要一个disease code,不一定是1/0,但最后选择编码为1/0
如不1/0,则stat_roc默认按顺序最低值为无病状态
basicplot <- ggplot(test, aes(d = D, m = M1)) + geom_roc()
basicplot
若diseaase编码非1/0: 提示warning但仍能继续
ggplot(test, aes(d = D.str, m = M1)) + geom_roc()
n.cuts参数:展示几个截断点
labelsize: 展示标签的大小
labelround: label值保留几位小数
ggplot(test, aes(d = D, m = M1)) + geom_roc(n.cuts = 5, labelsize = 5, labelround = 2)
修改style-style_roc函数
styledplot <- basicplot + style_roc()
styledplot
修改xlab, 主题
basicplot + style_roc(theme = theme_grey, xlab = "1 - Specificity")
multiROC-多因素诊断
meltroc类似于dplyr中的gather。转换数据为长数据,原数据为两列marker
head(test)
## D M name
## M11 1 1.48117155 M1
## M12 1 0.61994478 M1
## M13 0 0.57613345 M1
## M14 1 0.85433197 M1
## M15 0 0.05258342 M1
## M16 1 0.66703989 M1
longtest <- melt_roc(test, "D", c("M1", "M2"))
head(longtest)
table(longtest$name)
## ROC曲线比较
ggplot(longtest, aes(d = D, m = M, color = name)) +
geom_roc() +
style_roc()+
ggsci::scale_color_lancet()
ggplot2分面
ggplot(longtest, aes(d = D, m = M, color = name)) +
geom_roc() +
style_roc()+
facet_wrap(~name)+
ggsci::scale_color_lancet()
主题与注释
AUC计算并绘制在图中-calc_auc函数
calc_auc(basicplot)$AUC提取
basicplot +
style_roc(theme = theme_grey) + ##主题修改
theme(axis.text = element_text(colour = "blue")) +
ggtitle("Themes and annotations") + ## 标题
annotate("text", x = .75, y = .25, ## 注释text的位置
label = paste("AUC =", round(calc_auc(basicplot)$AUC, 2))) +
scale_x_continuous("1 - Specificity", breaks = seq(0, 1, by = .1)) ## x刻度
## Scale for 'x' is already present. Adding another scale for 'x', whi
对multi_ROC注释,实现多个AUC值的呈现, 实际上仍然是ggplot2语法中的annotate注释
p<-ggplot(longtest, aes(d = D, m = M, color = name)) +
geom_roc(n.cuts = 0) +
style_roc()+
ggsci::scale_color_lancet()
auc<-calc_auc(p)
head(auc)
## PANEL group AUC
## 1 1 1 0.833985
## 2 1 2 0.679599
p+annotate("text",x = .75, y = .25, ## 注释text的位置
label = paste("AUC of M1 =", round(calc_auc(p)$AUC[1], 2))) +
annotate("text",x = .75, y = .15, ## 注释text的位置)
label=paste("AUC of M2 =", round(calc_auc(p)$AUC[2], 2)))
其它计算ROC曲线的算法融入
默认的calculate_roc 计算的是 empirical ROC曲线
只要有cutoff, TPF,FPF即可计算,将这些结果以数据框的形式传入到 ggroc 函数
代替默认的统计方法为identity
require(plotROC)
require(ggplot2)
set.seed(2529)
D.ex <- rbinom(200, size = 1, prob = .5)
M1 <- rnorm(200, mean = D.ex, sd = .65)
M2 <- rnorm(200, mean = D.ex, sd = 1.5)
test <- data.frame(D = D.ex, D.str = c("Healthy", "Ill")[D.ex + 1],
M1 = M1, M2 = M2, stringsAsFactors = FALSE)
head(test)
## D D.str M1 M2
## 1 1 Ill 1.48117155 -2.50636605
## 2 1 Ill 0.61994478 1.46861033
## 3 0 Healthy 0.57613345 0.07532573
## 4 1 Ill 0.85433197 2.41997703
## 5 0 Healthy 0.05258342 0.01863718
## 6 1 Ill 0.66703989 0.24732453
D.ex <- test$D
M.ex <- test$M1
mu1 <- mean(M.ex[D.ex == 1])
mu0 <- mean(M.ex[D.ex == 0])
s1 <- sd(M.ex[D.ex == 1])
s0 <- sd(M.ex[D.ex == 0])
c.ex <- seq(min(M.ex), max(M.ex), length.out = 300)
## 构造数据框传入数据
binorm.roc <- data.frame(c = c.ex,
FPF = pnorm((mu0 - c.ex)/s0),
TPF = pnorm((mu1 - c.ex)/s1)
)
head(binorm.roc)
binorm.plot <- ggplot(binorm.roc, aes(x = FPF, y = TPF, label = c)) +
geom_roc(stat = "identity") + style_roc(theme = theme_grey)
binorm.plot
时间依赖的ROC曲线
配合survival ROC包
配合lapply函数实现批量绘图
lappy的结果返回为list,刚好输入do.call
require(ggplot2)
require(plotROC)
library(survivalROC)
survT <- rexp(350, 1/5)
cens <- rbinom(350, 1, .1)
M <- -8 * sqrt(survT) + rnorm(350, sd = survT)
### 时间2,5,10
sroc <- lapply(c(2, 5, 10), function(t){
stroc <- survivalROC(Stime = survT, status = cens, marker = M,
predict.time = t, method = "NNE", ## KM法或NNE法
span = .25 * 350^(-.2))
data.frame(TPF = stroc[["TP"]], FPF = stroc[["FP"]],
c = stroc[["cut.values"]],
time = rep(stroc[["predict.time"]], length(stroc[["FP"]])))
})
## 整合到数据框中
sroclong <- do.call(rbind, sroc)
class(sroclong)
## [1] "data.frame"
head(sroclong)
## TPF FPF c time
## 1 1 1.0000000 -Inf 2
## 2 1 0.9970286 -96.21091 2
## 3 1 0.9940573 -89.13315 2
## 4 1 0.9910859 -80.53402 2
## 5 1 0.9881145 -70.53104 2
## 6 1 0.9851431 -67.81392 2
sroclong$time<-factor(sroclong$time)
## 绘制ROC
pROC<-ggplot(sroclong, aes(x = FPF, y = TPF, label = c, color = time)) +
geom_roc(labels = FALSE, stat = "identity") +
style_roc()+
ggsci::scale_color_jco()
pROC
添加注释
pROC+annotate("text",x = .75, y = .25, ## 注释text的位置
label = paste("AUC of 1 years =", round(calc_auc(pROC)$AUC[1], 2))) +
annotate("text",x = .75, y = .15, ## 注释text的位置)
label=paste("AUC of 3 years =", round(calc_auc(pROC)$AUC[2], 2)))+
annotate("text",x = .75, y = .05, ## 注释text的位置)
label=paste("AUC of 5 years =", round(calc_auc(pROC)$AUC[3], 2)))