Retrive METABRIC data from cBioPortal

cbioportal-metabric.R

library(dplyr)
library(tidyr)
library(ggplot2)
library(cgdsr)

ggplot2::theme_set(theme_classic())

conn <- CGDS("http://www.cbioportal.org/")

test(conn)

## Inital look at availability
x1 <- getCancerStudies(conn)
rownames(x1) <- x1$cancer_study_id
x1[, "name", drop = FALSE]
subset(x1, grepl("brca", cancer_study_id), "name")

x2 <- getCaseLists(conn, "brca_metabric")
x2$case_list_id

x3 <- getGeneticProfiles(conn, "brca_metabric")
x3$genetic_profile_id

## Retrieve data from cBioPortal
df_clin <- getClinicalData(conn,
  caseList = "brca_metabric_all")
df_clin$subjid <- row.names(df_clin)

df_mrna <- getProfileData(conn,
  genes = c("ESR1", "ERBB2"),
  geneticProfiles = "brca_metabric_mrna",
  caseList = "brca_metabric_all")
names(df_mrna) <- paste0(names(df_mrna), "_mrna")
df_mrna$subjid <- row.names(df_mrna)

df_cna <- getProfileData(conn,
  genes = c("ESR1", "ERBB2"),
  geneticProfiles = "brca_metabric_cna",
  caseList = "brca_metabric_all")
names(df_cna) <- paste0(names(df_cna), "_cna")
df_cna$subjid <- row.names(df_cna)

df <- df_clin %>%
  left_join(df_mrna, by = "subjid") %>%
  left_join(df_cna, by = "subjid")

colcna <- c(rev(RColorBrewer::brewer.pal(5, "RdYlBu")), "grey")
names(colcna) <- c(as.character(-2:2), "NaN")

## Look at data
summary(df)

gg1 <- df %>%
  ggplot(aes(x = THREEGENE, fill = factor(ERBB2_cna))) +
  geom_bar(position = "fill") +
  scale_fill_manual(values = colcna) +
  labs(y = "Proportion") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

gg2 <- df %>%
  ggplot(aes(x = factor(ERBB2_cna), y = ERBB2_mrna)) +
  geom_boxplot(aes(fill = factor(ERBB2_cna))) +
  scale_fill_manual(values = colcna, guide = FALSE)

cowplot::plot_grid(gg1, gg2, labels = "AUTO")