title	author	date	vignette	output
patelGBMSC -- CONQUER quantification of single-cell RNA-seq in glioblastoma, thinned, with colData	Vincent J. Carey, stvjc at channing.harvard.edu	`r format(Sys.time(), '%B %d, %Y')`	%\VignetteEngine{knitr::rmarkdown} %\VignetteIndexEntry{patelGBMSC -- a single-cell RNA-seq dataset in glioblastoma} %\VignetteEncoding{UTF-8}	BiocStyle::pdf_document BiocStyle::html_document toc number_sections true true highlight number_sections theme toc pygments true united true

suppressPackageStartupMessages({
suppressMessages({
library(patelGBMSC)
})
})

Introduction

Patel et al. 2014 describe a single-cell RNA-seq study of several glioblastoma samples. The data were reprocessed with the CONQUER pipeline (see the QC report).

The rds file distributed by CONQUER is large as it includes multiple gene-level and transcript-level quantifications. As of Oct 30 2017, the CONQUER distribution does not include sample-level information beyond the GSM identifier. This package includes a smaller image of the data (the count_lstpm quantifications, that are estimated counts created using the salmon algorithm, rescaled to account for library size). The data image is over 200MB, so the r Biocpkg("BiocFileCache") discipline is used to perform a one-time download, insertion and bookkeeping in cache; the loadPatel function takes care of the download and retrieval from cache as appropriate.

Quick view of the data

We'll randomly sample 5000 genes to reduce runtime in this vignette. We filter down to the 430 patient samples that passed quality control.

library(patelGBMSC)
patelGeneCount = loadPatel()
qdrop = grep("excluded", patelGeneCount$description) # QC issues
patelGeneCount = patelGeneCount[,-qdrop]
ispat = grep("MGH", patelGeneCount$characteristics_ch1)
patelGeneCount = patelGeneCount[,ispat]
patelGeneCount = patelGeneCount[-grep("ERCC", rownames(patelGeneCount)),] # drop ERCC spikeins
patelGeneCount$sampcode = factor(gsub("patient id: ", "", patelGeneCount$characteristics_ch1))
tcol = as.numeric(tfac <- factor(patelGeneCount$sampcode))
set.seed(1234)
samp = assay(patelGeneCount[sample(1:nrow(patelGeneCount), size=5000),])
library(Rtsne)
RTL = Rtsne(t(log(samp+1)))
#plot(RTL$Y, col=tcol, pch=19)
#legend(8,15,legend=levels(tfac),col=1:length(levels(tfac)),
  # pch=19)
myd = data.frame(ts1=RTL$Y[,1], ts2=RTL$Y[,2], 
        code = patelGeneCount$sampcode, tcol=tcol)
library(ggplot2)
ggplot(myd, aes(x=ts1, y=ts2, group=code, colour=code)) + geom_point()