README.md

In vjcitn/patelGBMSC: A SummarizedExperiment instance with single-cell RNA-seq on glioblastoma samples

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

```{r setup,echo=FALSE,results="hide"} suppressPackageStartupMessages({ suppressMessages({ library(patelGBMSC) }) })

# Introduction

[Patel et al. 2014](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4123637/)
describe a single-cell RNA-seq study of several glioblastoma
samples.  The data were reprocessed with the 
[CONQUER](http://imlspenticton.uzh.ch:3838/conquer/)
pipeline (see the [QC report](http://imlspenticton.uzh.ch/robinson_lab/conquer/report-multiqc/GSE57872_multiqc_report.html)).

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.

```{r getdat}
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()

vjcitn/patelGBMSC documentation built on May 14, 2019, 12:48 p.m.