## ROC曲線 ### 應用場景 評估某個指標預測某項事件的性能，診斷的特異性和靈敏性，比如以基因的表達預測tumor和normal的特異性和靈敏性 > 使用數據格式如下圖  ```R load('./roc_test.Rdata') #rt ### 計算給予TP53表達判斷tumor和normal的特異性和靈敏性 roc_pt <- roc(rt_roc$group, rt_roc$TP53) AUC_val <- round(as.numeric(roc_pt$auc), digits = 3)#獲取AUC的值 mat_roc <- data.frame(cbind(roc_pt$specificities, roc_pt$sensitivities), stringsAsFactors = FALSE) colnames(mat_roc) <- c('specificities', 'sensitivities') ``` > 作圖 ```R library(ggplot2) library(pROC) ggplot(mat_roc, aes(1- specificities, sensitivities)) + geom_point() + geom_abline(linetype="dashed", color = "grey") + annotate(geom="text", x = 0.75, y = 0.25, label = paste ('AUC = ' , AUC_val, sep = "" )) + theme_bw() # 將點換成線 # geom_line()做切換 ggplot(mat_roc, aes(1- specificities, sensitivities)) + geom_line(size = 1.5, colour = 'blue') + geom_abline(linetype="dashed", color = "grey") + annotate(geom="text", x = 0.75, y = 0.25, label = paste ('AUC = ' , AUC_val, sep = "" )) + theme_bw() ```