Question

RTCGA survival package Breast cancer subtypes

1

Entering edit mode

giulia_m ▴ 10

@giulia_m-16220

Last seen 6.4 years ago

Hello everyone!

I used the RTCGA package for survival analyis in different types of cancers from TCGA. I stratified the patients based on high or low expression of two genes. By using the code generated by MarcinKosinski I obtained nice survival curves (https://github.com/RTCGA/RTCGA/issues/97#categorize-variables). The R code he suggests is immediate and very useful!

Now I would like to stratify the patients based on the different breast cancer subtypes (Luminal A, Luminal B etc....) to see if the genes I considered affect the survival of particular subtypes.

Do you have any suggestions for this?

How to get clinical and RNA data from each subtype?

Best,

Giulia

rtcga • 1.8k views

ADD COMMENT • link 6.4 years ago giulia_m ▴ 10

0

Entering edit mode

Hello @giulia_m , can you provide the code, so we can update it with your request?

ADD REPLY • link 6.4 years ago MarcinKosiński • 0

0

Entering edit mode

Here is the code I used:

library(RTCGA.clinical)
library(RTCGA.rnaseq)
library(tidyverse)
library(survmnier)

survivalTCGA(BRCA.clinical) -> BRCA.surv
expressionsTCGA(
BRCA.rnaseq,
extract.cols = c("CDCA2|157313", "AURKB|9212")
) -> BRCA.rnaseq

dim(BRCA.surv); dim(BRCA.rnaseq)

head(BRCA.surv); head(BRCA.rnaseq)

BRCA.rnaseq <- BRCA.rnaseq %>%
rename(cohort = dataset,
CDCA2 = `CDCA2|157313`,
AURKB = `AURKB|9212`) %>%
filter(substr(bcr_patient_barcode, 14, 15) == "01") %>%
mutate(bcr_patient_barcode = substr(bcr_patient_barcode, 1, 12))

head(BRCA.rnaseq)

BRCA.surv %>%
left_join(BRCA.rnaseq,
by = "bcr_patient_barcode") ->
BRCA.surv_rnaseq

head(BRCA.surv_rnaseq)

table(BRCA.surv_rnaseq$cohort, useNA = "always")

BRCA.surv_rnaseq <- BRCA.surv_rnaseq %>%
filter(!is.na(cohort))

dim(BRCA.surv_rnaseq)

BRCA.surv_rnaseq.cut <- surv_cutpoint(
BRCA.surv_rnaseq,
time = "times",
event = "patient.vital_status",
variables = c("CDCA2", "AURKB")
)
summary(BRCA.surv_rnaseq.cut)

BRCA.surv %>%
left_join(BRCA.rnaseq,
by = "bcr_patient_barcode") ->
BRCA.surv_rnaseq

head(BRCA.surv_rnaseq)

BRCA.surv_rnaseq.cat <- surv_categorize(BRCA.surv_rnaseq.cut)
headBRCA.surv_rnaseq.cat)

BRCA.surv_rnaseq.cat <- BRCA.surv_rnaseq.cat %>%
mutate(cohort = BRCA.surv_rnaseq$cohort)
headBRCA.surv_rnaseq.cat)

library(survival)
fit <- survfit(Surv(times, patient.vital_status) ~ CDCA2 + AURKB,
data = BRCA.surv_rnaseq.cat)

Thanks!

ADD REPLY • link 6.4 years ago giulia_m ▴ 10

0

Entering edit mode

@giulia_m is any column name meaningful if you run colnames(BRCA.clinical) then you can specify extract.cols in survivalTCGA . You can try using


lapply(BRCA.clinical, function(x) {any(grepl('lum', tolower(x)))}) -&gt; is_there_luminal

which(unlist(is_there_luminal))

do detect whether any variables has lum in any name of considered factor levels

table(BRCA.clinical$patient.breast_carcinoma_primary_surgical_procedure_name)
                 lumpectomy              mastectomy nos modified radical mastectomy                       other 
                          9                          18                          14                          14

however I can't find luminal

ADD REPLY • link 6.4 years ago MarcinKosiński • 0

0

Entering edit mode

Thank you Marcin!

Yes I tried...

I cannot find any indication with "subtypes"...

ADD REPLY • link 6.4 years ago giulia_m ▴ 10

0

Entering edit mode

I could only check via: http://tumorsurvival.org/TCGA/Breast_TCGA_BRCA/index.html

Here you can select Luminal A, Luminal B etc.

For example here are the results: http://tumorsurvival.org/TCGA/Breast_TCGA_BRCA/process.php

It is possible to download the excel file but then how should I deal with?

ADD REPLY • link 6.4 years ago giulia_m ▴ 10

0

Entering edit mode

Is there a way to download sub-type dependency with the bcr_patient_barcode ? If yes, then one can join this to BRCA.clinical with dplyr::left_join

ADD REPLY • link 6.4 years ago MarcinKosiński • 0