专栏首页生信技能树寻找生存分析的最佳基因表达分组阈值

寻找生存分析的最佳基因表达分组阈值

文字版是:

Based on the FPKM value of each gene, we classified the patients into two groups and examined their prognoses. In the analysis, we excluded genes with low expression, i.e., those with a median expression among samples less than FPKM 1. The prognosis of each group of patients was examined by Kaplan-Meier survival estimators, and the survival outcomes of the two groups were compared by log-rank tests. To choose the best FPKM cut-offs for grouping the patients most significantly, all FPKM values from the 20th to 80th percentiles were used to group the patients, significant differences in the survival outcomes of the groups were examined and the value yielding the lowest log-rank P value is selected.

得到的K-M图如下:

image-20190516092236980

如果是在 http://www.oncolnc.org/ 出图如下:

a=read.csv('BRCA_10384_50_50.csv')
head(a)
a$event=ifelse(a$Status=='Alive',0,1)
library(survival)
library(survminer)
sfit <- survfit(Surv(Days, event)~Group, data=a) 
ggsurvplot(sfit, conf.int=F, pval=TRUE)

phe=a
phe$time=phe$Days/365

## 批量生存分析 使用  logrank test 方法
mySurv=with(phe,Surv(time, event))
log_rank_p <- lapply(2:8, function(i){
  thr=sort(phe$Expression)[round(nrow(phe)*i/10)]
  phe$group=ifelse(phe$Expression > thr,'high','low') 
  print(table( phe$group ))
  data.survdiff=survdiff(mySurv~group,data=phe)
  p.val = 1 - pchisq(data.survdiff$chisq, length(data.survdiff$n) - 1)
  return(p.val)
}) 
log_rank_p=unlist(log_rank_p)
log_rank_p

i=8
thr=sort(phe$Expression)[round(nrow(phe)*i/10)]
phe$group=ifelse(phe$Expression > thr,'high','low') 
print(table( phe$group ))
sfit <- survfit(Surv(time, event)~group, data=phe) 
ggsurvplot(sfit, conf.int=F, pval=TRUE)

本文分享自微信公众号 - 生信技能树(biotrainee),作者:生信技能树

原文出处及转载信息见文内详细说明,如有侵权,请联系 yunjia_community@tencent.com 删除。

原始发表时间:2019-05-16

本文参与腾讯云自媒体分享计划,欢迎正在阅读的你也加入,一起分享。

我来说两句

0 条评论
登录 后参与评论

相关文章

  • WashU EpiGenome Browser使用教程

    WashU EpiGenome Browser 是我用过最赞的浏览器,没有之一。希望大伙跟着教程好好学习下! 还有更多教程见:http://epigenomeg...

    生信技能树
  • 这可能是我见过最简单的一篇SCI了

    批次很多时候无法避免,比如文章 Biomed Res Int. 2014 . doi: 10.1155/2014/319534 就提到:

    生信技能树
  • 使用IMonitor进行免疫组库分析

    其实前面我们已经分享了MiXCR,还有igblast,这两个免疫组库上游分析软件已经够用,如下:

    生信技能树
  • No thread-bound request found异常

    本文主要研究下spring mvc的No thread-bound request found异常

    codecraft
  • Head First Stanford NLP (1)

    (深入浅出Stanford NLP 基础篇) 本文主要介绍Stanford NLP工具的基本使用方法。

    宅男潇涧
  • How to build Multi-Language Web Sites with ASP.NET 2.0 and VS.Net 2005

    Introduction: In order to reach international markets through the Internet, sup...

    张善友
  • Golang memory model

    Introduction The Go memory model specifies the conditions under which reads of a...

    李海彬
  • Create a natural language classifier that identifies spam

    With the advent of cognitive computing and smart machines, machine learning and ...

    首席架构师智库
  • 利用setTimeout和SetInterval构建Javascript计时器

    看到了一篇深入浅出的讲解setTimeout和setInterval的例子,直接讲英文贴出来吧,也不是很难。

    大江小浪
  • 超弱局部间断伽辽金法研究具有高阶空间导数的偏微分方程(CS NA)

    本文提出了一种新的不连续伽辽金方法,用于求解具有高阶空间导数的几类偏微分方程。将局部间断伽辽金法和超弱间断伽辽金法的优点结合起来。首先将具有高阶空间导数的偏微分...

    非过度曝光

扫码关注云+社区

领取腾讯云代金券