Question: cBioportal survival analysis help
0
gravatar for mfangli
2.4 years ago by
mfangli0
mfangli0 wrote:

hello, every one! I am a newcomer. Recently, I try to use cBioportal to analyzie survival. For example, I want to find whether tumor sample expression of IL8,CXCR1,CXCR2 genes can infuence breast cancer patients prognosis. Select TCGA, Nature Communications 2016, mRNA microarray, z-score ± 2.00. Input IL8,CXCR1,CXCR2 gene list menu. The result show us that tumor sample expression of IL8,CXCR1,CXCR2 genes were stratified into 3 groups(upregulation, downregulation ,noalterlation). But the survival analysis is based on two groups (noalterlation,alterlation).The alterlation group should include upregulation and downregulation.If I want to compare upregulation group with noalterlation group, how shuould I do ?

rna-seq gene • 2.5k views
ADD COMMENTlink modified 2.4 years ago by TriS4.0k • written 2.4 years ago by mfangli0
1
gravatar for Samuel Brady
2.4 years ago by
Samuel Brady300
Samuel Brady300 wrote:

Unfortunately, there is no way to do this in cBioPortal. cBioPortal will call patients with either up-regulation OR down-regulation "altered," while those with a z-score between -2 and 2 "non-altered." This of course makes no sense.

If you'd like something simple, I'd use KM plotter. If you want more freedom you can download TCGA RNA-Seq and survival data from GDC Data Portal.

ADD COMMENTlink modified 2.4 years ago • written 2.4 years ago by Samuel Brady300
1
gravatar for TriS
2.4 years ago by
TriS4.0k
United States, Buffalo
TriS4.0k wrote:

if you have some R skills a way to do that is to calculate the "ntile" per each of those genes (i.e. top vs. bottom quartile/tercile etc..) and then compare patients that have high vs. low expression.

for exemple (in pseudocode):

# take quantiles high and low 
gene <-  expression levels of gene X 
g_high <- take top quartile 
g_low <- take low quartile

# define a class for the patients with high/low expression 
class <- if expression of X is >= g_high <- 1
      else
       if expression of X is <= g_low <- 2
     else
       class <- 0
 # keep only high and low, remove those in between 
ind_to_remove <- which class == 0 
alive_dead <- if dead <- 1 else <- 0

 # prepare your formula for analysis
myAnalysis <- surv(OS[-ind_to_remove], alive_dead[-ind_to_remove]) ~ class[-ind_to_remove]
 # run logrank test 
s_fit <- survfit(myAnalysis) 
s_diff <- survdiff(myAnalysis)
 # and cox hazard model 
cox <- coxph(myAnalysis)

 # plot KM 
plot(s)

you can always keep those in between (i.e. class == 0), up to you what to do with them

also, check this: Survival analysis of TCGA patients integrating gene expression (RNASeq) data

ADD COMMENTlink modified 2.4 years ago • written 2.4 years ago by TriS4.0k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1191 users visited in the last hour