In lootpiz/rePurIST: Implementation of PurIST

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Introduction

Implementation of Purity Independent Subtyping of Tumors (PurIST) that was published in the Clinical Cancer Research (Rashid and Peng et al, 2020; DOI: 10.1158/1078-0432.CCR-19-1467).

Install the package

intall.packages("devtools")
library(devtools)
devtools::install_github("lootpiz/PurIST")

Load the package

library(PurIST)

A single sample classicifaction

prob = PurIST(dat)

MetaGxPancreas subtyping

```r library(PurIST) library(MetaGxPancreas) library(data.table)

esets <- MetaGxPancreas::loadPancreasEsets()[[1]]

esetExprs <- lapply(esets, FUN=exprs) esetNames <- names(esetExprs) esetSamples <- lapply(esetExprs, FUN=colnames)

.tryPurIST <- function(col, names) { tryCatch({PurIST(structure(col, .Names=names))}, error=function(e) { return(NA) }) }

.PurISTExprColumns <- function(expr) apply(expr, 2, FUN=.tryPurIST, names=row.names(expr))

objList <- lapply(esetExprs, FUN=.PurISTExprColumns)

resultsDtL <- mapply(data.table, 'study'=esetNames, 'sample'=esetSamples, 'prob'=objList, SIMPLIFY=FALSE) resultsDT <- rbindlist(resultsDtL) resultsDT[, subtype := ifelse(prob > 0.5, "Basal-like", "Classical")]

table(resultsDT$subtype)

lootpiz/rePurIST documentation built on July 18, 2020, 7:03 p.m.