DeLorean analysis of Trapnell et al. myoblast data

In DeLorean: Estimates Pseudotimes for Single Cell Expression Data

devtools::load_all('../..')
devtools::load_all('../../../DeLoreanData')
rmarkdown::render('DeLorean-Trapnell.Rmd')

library(knitr)
library(knitcitations)
library(rmarkdown)
#
# knitr options
#
opts_chunk$set(
    fig.path = 'figures/Trapnell-',
    stop_on_error = TRUE,
    fig.width = 12.5,
    fig.height = 8)
#
# Citations
#
cleanbib()
cite_options(
    # hyperlink = 'to.doc',
    hyperlink = TRUE,
    # style = 'html',
    # citation_format = 'text',
    citation_format = "pandoc",
    cite.style = "numeric",
    check.entries = TRUE)
    # hyperlink = TRUE)
bib <- read.bibtex("DeLorean.bib")
if (file.exists("config.R")) {
    source("config.R")
}

# suppressMessages(loadfonts())
library(DeLorean)
library(DeLoreanData)
#
# Stylesheet
#
options(markdown.HTML.stylesheet = system.file(file.path('Rmd', 'foghorn.css'),
                                               package="DeLorean"))

Load data.

library(DeLorean)
data(TrapnellDeLorean)
dl <- de.lorean(
    trapnell.expr,
    trapnell.gene.meta,
    trapnell.cell.meta)

Estimate hyperparameters

Examine data for empirical Bayes estimation of hyperparameters.

dl <- estimate.hyper(dl, sigma.tau=6)

Choose genes and cells

Select cells at random if we have too many.

set.seed(1)
max.cells <- min(getOption("DL.max.cells", 57))
sampled.cells <- sample(dl$cell.meta$cell, max.cells)
cell.filter <- function(cells) cells %in% sampled.cells
dl <- filter_cells(dl, cell.filter)

Select genes by variance.

max.genes <- 151
genes.high.var <- (dl$gene.var
                   %>% arrange(-Psi)
                   %>% head(max.genes))$gene
gene.filter <- function(genes) genes %in% genes.high.var
dl <- filter_genes(dl, gene.filter)

Compile and fit model

Define and compile the model, find the best initialisation, and fit the model.

dl <- prepare.for.stan(dl)
dl <- compile.model(dl)
dl <- find.best.tau(dl)
system.time(dl <- fit.model(dl))

Examine convergence.

dl <- examine.convergence(dl)

Analyse posterior

Examine posterior.

dl <- process.posterior(dl)
dl <- analyse.noise.levels(dl)

Profiles

Calculate expression profiles.

dl <- make.predictions(dl)

Plot some specific profiles mentioned in Trapnell et al.

genes.fig.2 <- c("MEF2C", "MYH2", "CDK1", "ID1", "MYOG")
genes.fig.3.cluster <- c("HES1", "PBX1", "MYOG", "CDK1", "AKT1", "MYF6")
genes.fig.3.reg <- c("MEF2", "MYOD", "P300", "E47", "MEIS1", "PBX1",
                     "XBP1", "CUX1", "USF1", "ZIC1", "MZF1", "POU2F1",
                     "POU3F2", "AHR", "RREB1", "RFX1", "HIVEP2", "PATZ1",
                     "DDIT3", "BACHA1", "ARID5B")
genes.fig.3 <- c(genes.fig.3.cluster, genes.fig.3.reg)
genes.fig.4.a <- c("MZF1", "ZIC1", "XBP1", "CUX1", "ARID5B", "USF1", "POU2F1",
                   "AHR", "PATZ1", "DDIT3", "BACH1", "MYOG")
genes.fig.all <- unique(c(genes.fig.2, genes.fig.3, genes.fig.4.a))
filter(dl$gene.map , hgnc_symbol %in% genes.fig.all)
filter(dl$gene.meta, hgnc_symbol %in% genes.fig.all)
png('Trapnell-profiles.png', width=1200, height=800)
(plot(dl,
      type="profiles",
      genes=filter(dl$gene.map, hgnc_symbol %in% genes.fig.all)$gene)
    + facet_wrap(~ hgnc_symbol + gene))
dev.off()

# Save DeLorean object without fit component
saveRDS({dl2 <- dl; dl2$fit <- NULL; dl2}, "Data/Trapnell.rds")

R version and packages used:

sessionInfo()

DeLorean documentation built on May 2, 2019, 9:24 a.m.