From 8b98f4acb5258df432119a50a9ed7dec564759a5 Mon Sep 17 00:00:00 2001
From: Luke Zappia <luke.zappia@mcri.edu.au>
Date: Sun, 16 Jul 2017 06:30:22 +0000
Subject: [PATCH] Merge branch 'master' into devel
* master:
Bump version number
Run checks
Patch fix to lun2Estimate from basics branch
Update summariseDiff
Add counts check to scDDEstimate
Adjust nEE/nEP estimation
Update scDD estimation
From: Luke Zappia <lazappi@users.noreply.github.com>
git-svn-id: file:///home/git/hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/splatter@131258 bc3139a8-67e5-0310-9ffc-ced21a209358
---
DESCRIPTION | 11 ++--
NEWS.md | 7 +++
R/AllClasses.R | 18 +++---
R/SCDDParams-methods.R | 12 ++--
R/compare.R | 129 ++++++++++++++++++++++++++++++++++++-----
R/lun2-estimate.R | 12 +++-
R/scDD-estimate.R | 52 ++++++++++++++---
man/SCDDParams.Rd | 18 +++---
man/scDDEstimate.Rd | 22 +++++--
man/summariseDiff.Rd | 11 ++--
10 files changed, 225 insertions(+), 67 deletions(-)
diff --git a/DESCRIPTION b/DESCRIPTION
index f715be4..594612b 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -1,8 +1,8 @@
Package: splatter
Type: Package
Title: Simple Simulation of Single-cell RNA Sequencing Data
-Version: 1.1.1
-Date: 2017-07-07
+Version: 1.1.2
+Date: 2017-07-16
Author: Luke Zappia
Authors@R:
c(person("Luke", "Zappia", role = c("aut", "cre"),
@@ -35,7 +35,8 @@ Imports:
matrixStats,
ggplot2,
scales,
- BiocParallel
+ BiocParallel,
+ SummarizedExperiment
Suggests:
testthat,
scran,
@@ -48,8 +49,8 @@ Suggests:
BiocStyle,
covr,
S4Vectors,
- SummarizedExperiment,
- cowplot
+ cowplot,
+ limSolve
biocViews: SingleCell, RNASeq, Transcriptomics, GeneExpression, Sequencing,
Software
URL: https://github.com/Oshlack/splatter
diff --git a/NEWS.md b/NEWS.md
index 7cb1f83..3ced178 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -1,3 +1,10 @@
+# 1.1.2
+
+* Update summariseDiff
+* Update scDDEstimate, now estimates gene types
+* Fix error in lun2Estimate
+* Import SummarizedExperiment to avoid warnings
+
# 1.1.1
* Add batch effects to Splat simulation
diff --git a/R/AllClasses.R b/R/AllClasses.R
index 3b37656..59432fe 100644
--- a/R/AllClasses.R
+++ b/R/AllClasses.R
@@ -405,24 +405,24 @@ setClass("Lun2Params",
#' The SCDD simulation uses the following parameters:
#'
#' \describe{
-#' \item{\code{[nGenes]}}{The number of genes to simulate (not used).}
+#' \item{\code{nGenes}}{The number of genes to simulate (not used).}
#' \item{\code{nCells}}{The number of cells to simulate in each condition.}
#' \item{\code{[seed]}}{Seed to use for generating random numbers.}
#' \item{\code{SCdat}}{
#' \code{\link[SummarizedExperiment]{SummarizedExperiment}} containing real
#' data.}
-#' \item{\code{[nDE]}}{Number of DE genes to simulate.}
-#' \item{\code{[nDP]}}{Number of DP genes to simulate.}
-#' \item{\code{[nDM]}}{Number of DM genes to simulate.}
-#' \item{\code{[nDB]}}{Number of DB genes to simulate.}
-#' \item{\code{[nEE]}}{Number of EE genes to simulate.}
-#' \item{\code{[nEP]}}{Number of EP genes to simulate.}
+#' \item{\code{nDE}}{Number of DE genes to simulate.}
+#' \item{\code{nDP}}{Number of DP genes to simulate.}
+#' \item{\code{nDM}}{Number of DM genes to simulate.}
+#' \item{\code{nDB}}{Number of DB genes to simulate.}
+#' \item{\code{nEE}}{Number of EE genes to simulate.}
+#' \item{\code{nEP}}{Number of EP genes to simulate.}
#' \item{\code{[sd.range]}}{Interval for fold change standard deviations.}
#' \item{\code{[modeFC]}}{Values for DP, DM and DB mode fold changes.}
#' \item{\code{[varInflation]}}{Variance inflation factors for each
-#' condition. If all equal to 1 will be set to \code{NULL} (default)}
+#' condition. If all equal to 1 will be set to \code{NULL} (default).}
#' \item{\code{[condition]}}{String giving the column that represents
-#' biological group of interest}
+#' biological group of interest.}
#' }
#'
#' The parameters not shown in brackets can be estimated from real data using
diff --git a/R/SCDDParams-methods.R b/R/SCDDParams-methods.R
index 97c839d..09aa786 100644
--- a/R/SCDDParams-methods.R
+++ b/R/SCDDParams-methods.R
@@ -86,12 +86,12 @@ setMethod("setParam", "SCDDParams", function(object, name, value) {
setMethod("show", "SCDDParams", function(object) {
- pp <- list("Genes:" = c("[nDE]" = "nDE",
- "[nDP]" = "nDP",
- "[nDM]" = "nDM",
- "[nDP]" = "nDP",
- "[nEE]" = "nEE",
- "[nEP]" = "nEP"),
+ pp <- list("Genes:" = c("(nDE)" = "nDE",
+ "(nDP)" = "nDP",
+ "(nDM)" = "nDM",
+ "(nDP)" = "nDP",
+ "(nEE)" = "nEE",
+ "(nEP)" = "nEP"),
"Fold change:" = c("[SD Range]" = "sd.range",
"[Mode FC]" = "modeFC"),
"Variance:" = c("[Inflation]" = "varInflation"),
diff --git a/R/compare.R b/R/compare.R
index f70c76b..ffee65d 100644
--- a/R/compare.R
+++ b/R/compare.R
@@ -790,20 +790,19 @@ makeOverallPanel <- function(comp, diff, title = "Overall comparison",
#' Summarise diffSCESets
#'
-#' Summarise the results of \code{\link{diffSCESets}}. The various
-#' properties are sorted, differences calculated, the Median Absolute Deviation
-#' taken as the summary statistic and the ranks calculated.
+#' Summarise the results of \code{\link{diffSCESets}}. Calculates the Median
+#' Absolute Deviation (MAD), Mean Absolute Error (MAE) and Root Mean Squared
+#' Error (RMSE) for the various properties and ranks them.
#'
#' @param diff Output from \code{\link{diffSCESets}}
#'
-#' @return List with MADs, ranks and both combined in long format
+#' @return data.frame with MADs, MAEs, RMSEs, scaled statistics and ranks
#' @examples
#' sim1 <- splatSimulate(nGenes = 1000, batchCells = 20)
#' sim2 <- simpleSimulate(nGenes = 1000, nCells = 20)
#' difference <- diffSCESets(list(Splat = sim1, Simple = sim2), ref = "Simple")
#' summary <- summariseDiff(difference)
-#' names(summary)
-#' head(summary$Long)
+#' head(summary)
#' @export
summariseDiff <- function(diff) {
@@ -819,6 +818,7 @@ summariseDiff <- function(diff) {
return(c(Mean = mean, Variance = var, ZerosGene = zeros,
MeanVar = mean.var, MeanZeros = mean.zeros))
})
+ fData.mads.z <- t(scale(t(fData.mads)))
pData.mads <- sapply(datasets, function(dataset) {
df <- diff$PhenoData[diff$PhenoData$Dataset == dataset, ]
@@ -826,27 +826,124 @@ summariseDiff <- function(diff) {
zeros <- median(abs(df$RankDiffZeros))
return(c(LibSize = lib.size, ZerosCell = zeros))
})
+ pData.mads.z <- t(scale(t(pData.mads)))
mads <- data.frame(Dataset = datasets, t(fData.mads), t(pData.mads))
+ mads.z <- data.frame(Dataset = datasets, t(fData.mads.z), t(pData.mads.z))
fData.ranks <- matrixStats::rowRanks(fData.mads)
pData.ranks <- matrixStats::rowRanks(pData.mads)
- ranks <- data.frame(Dataset = datasets, t(fData.ranks), t(pData.ranks))
- colnames(ranks) <- paste0(colnames(mads), "Rank")
+ ranks.mads <- data.frame(Dataset = datasets, t(fData.ranks), t(pData.ranks))
+ colnames(ranks.mads) <- paste0(colnames(mads), "Rank")
- mads.long <- stats::reshape(mads, varying = 2:8, direction = "long",
- idvar = "Dataset", timevar = "Statistic",
- times = colnames(mads)[2:8], v.names = "MAD")
+ fData.maes <- sapply(datasets, function(dataset) {
+ df <- diff$FeatureData[diff$FeatureData$Dataset == dataset, ]
+ mean <- mean(abs(df$RankDiffMeanLogCPM))
+ var <- mean(abs(df$RankDiffVarLogCPM))
+ zeros <- mean(abs(df$RankDiffZeros))
+ mean.var <- mean(abs(df$MeanRankVarDiff))
+ mean.zeros <- mean(abs(df$MeanRankZerosDiff))
+ return(c(Mean = mean, Variance = var, ZerosGene = zeros,
+ MeanVar = mean.var, MeanZeros = mean.zeros))
+ })
+ fData.maes.z <- t(scale(t(fData.maes)))
+
+ pData.maes <- sapply(datasets, function(dataset) {
+ df <- diff$PhenoData[diff$PhenoData$Dataset == dataset, ]
+ lib.size <- mean(abs(df$RankDiffLibSize))
+ zeros <- mean(abs(df$RankDiffZeros))
+ return(c(LibSize = lib.size, ZerosCell = zeros))
+ })
+ pData.maes.z <- t(scale(t(pData.maes)))
+
+ maes <- data.frame(Dataset = datasets, t(fData.maes), t(pData.maes))
+ maes.z <- data.frame(Dataset = datasets, t(fData.maes.z), t(pData.maes.z))
+
+ fData.ranks <- matrixStats::rowRanks(fData.maes)
+ pData.ranks <- matrixStats::rowRanks(pData.maes)
+
+ ranks.maes <- data.frame(Dataset = datasets, t(fData.ranks), t(pData.ranks))
+ colnames(ranks.maes) <- paste0(colnames(mads), "Rank")
+
+ fData.rmse <- sapply(datasets, function(dataset) {
+ df <- diff$FeatureData[diff$FeatureData$Dataset == dataset, ]
+ mean <- sqrt(mean(df$RankDiffMeanLogCPM ^ 2))
+ var <- sqrt(mean(df$RankDiffVarLogCPM ^ 2))
+ zeros <- sqrt(mean(df$RankDiffZeros ^ 2))
+ mean.var <- sqrt(mean(df$MeanRankVarDiff ^ 2))
+ mean.zeros <- sqrt(mean(df$MeanRankZerosDiff ^ 2))
+ return(c(Mean = mean, Variance = var, ZerosGene = zeros,
+ MeanVar = mean.var, MeanZeros = mean.zeros))
+ })
+ fData.rmse.z <- t(scale(t(fData.rmse)))
- ranks.long <- stats::reshape(ranks, varying = 2:8, direction = "long",
+ pData.rmse <- sapply(datasets, function(dataset) {
+ df <- diff$PhenoData[diff$PhenoData$Dataset == dataset, ]
+ lib.size <- sqrt(mean(df$RankDiffLibSize ^ 2))
+ zeros <- sqrt(mean(df$RankDiffZeros ^ 2))
+ return(c(LibSize = lib.size, ZerosCell = zeros))
+ })
+ pData.rmse.z <- t(scale(t(pData.rmse)))
+
+ rmse <- data.frame(Dataset = datasets, t(fData.rmse), t(pData.rmse))
+ rmse.z <- data.frame(Dataset = datasets, t(fData.rmse.z), t(pData.rmse.z))
+
+ fData.ranks <- matrixStats::rowRanks(fData.rmse)
+ pData.ranks <- matrixStats::rowRanks(pData.rmse)
+
+ ranks.rmse <- data.frame(Dataset = datasets, t(fData.ranks), t(pData.ranks))
+ colnames(ranks.rmse) <- paste0(colnames(rmse), "Rank")
+
+ mads <- stats::reshape(mads, varying = 2:8, direction = "long",
+ idvar = "Dataset", timevar = "Statistic",
+ times = colnames(mads)[2:8], v.names = "MAD")
+
+ mads.z <- stats::reshape(mads.z, varying = 2:8, direction = "long",
+ idvar = "Dataset", timevar = "Statistic",
+ times = colnames(mads)[2:8],
+ v.names = "MADScaled")
+
+ ranks.mads <- stats::reshape(ranks.mads, varying = 2:8, direction = "long",
idvar = "Dataset", timevar = "Statistic",
- times = colnames(ranks)[2:8], v.names = "Rank")
+ times = colnames(ranks.mads)[2:8],
+ v.names = "Rank")
+
+ maes <- stats::reshape(maes, varying = 2:8, direction = "long",
+ idvar = "Dataset", timevar = "Statistic",
+ times = colnames(maes)[2:8], v.names = "MAE")
- long <- data.frame(mads.long, Rank = ranks.long$Rank)
- row.names(long) <- NULL
+ maes.z <- stats::reshape(maes.z, varying = 2:8, direction = "long",
+ idvar = "Dataset", timevar = "Statistic",
+ times = colnames(mads)[2:8],
+ v.names = "MAEScaled")
+
+ ranks.maes <- stats::reshape(ranks.maes, varying = 2:8, direction = "long",
+ idvar = "Dataset", timevar = "Statistic",
+ times = colnames(ranks.maes)[2:8],
+ v.names = "Rank")
- summary <- list(MADs = mads, Ranks = ranks, Long = long)
+ rmse <- stats::reshape(rmse, varying = 2:8, direction = "long",
+ idvar = "Dataset", timevar = "Statistic",
+ times = colnames(mads)[2:8], v.names = "RMSE")
+
+ rmse.z <- stats::reshape(rmse.z, varying = 2:8, direction = "long",
+ idvar = "Dataset", timevar = "Statistic",
+ times = colnames(mads)[2:8],
+ v.names = "RMSEScaled")
+
+ ranks.rmse <- stats::reshape(ranks.rmse, varying = 2:8, direction = "long",
+ idvar = "Dataset", timevar = "Statistic",
+ times = colnames(ranks.rmse)[2:8],
+ v.names = "Rank")
+
+ summary <- data.frame(mads, MADScaled = mads.z$MADScaled,
+ MADRank = ranks.mads$Rank,
+ MAE = maes$MAE, MAEScaled = maes.z$MAEScaled,
+ MAERank = ranks.maes$Rank,
+ RMSE = rmse$RMSE, RMSEScaled = rmse.z$RMSEScaled,
+ RMSERank = ranks.rmse$Rank)
+ row.names(summary) <- NULL
return(summary)
}
diff --git a/R/lun2-estimate.R b/R/lun2-estimate.R
index ffd4155..cca9f6e 100644
--- a/R/lun2-estimate.R
+++ b/R/lun2-estimate.R
@@ -64,6 +64,11 @@ lun2Estimate.matrix <- function(counts, plates, params = newLun2Params(),
stop("The Lun2 simulation requires the 'pscl' package for estimation.")
}
+ if (!requireNamespace("limSolve", quietly = TRUE)) {
+ stop("The Lun2 simulation requires the 'limSolve' package for ",
+ "estimation.")
+ }
+
progress <- FALSE
if (requireNamespace("progress", quietly = TRUE)) {
progress <- TRUE
@@ -97,7 +102,12 @@ lun2Estimate.matrix <- function(counts, plates, params = newLun2Params(),
sizes <- seq(20, 100, 20)
}
sum.facs <- scran::computeSumFactors(dge$counts, cluster = groups,
- sizes = sizes)
+ sizes = sizes, positive = TRUE)
+ if (any(sum.facs == 0)) {
+ warning("Some sum factors are zero. See ?scran::computeSumFactors ",
+ "for details.")
+ sum.facs <- sum.facs + 1e-6
+ }
dge$samples$norm.factors <- sum.facs / dge$samples$lib.size
# Mean centre normalisation factors
dge$samples$norm.factors <- dge$samples$norm.factors /
diff --git a/R/scDD-estimate.R b/R/scDD-estimate.R
index 8f9a5c2..542d3af 100644
--- a/R/scDD-estimate.R
+++ b/R/scDD-estimate.R
@@ -7,27 +7,36 @@
#' @param conditions Vector giving the condition that each cell belongs to.
#' Conditions can be 1 or 2.
#' @param params SCDDParams object to store estimated values in.
+#' @param BPPARAM A \code{\link[BiocParallel]{BiocParallelParam}} instance
+#' giving the parallel back-end to be used. Default is
+#' \code{\link[BiocParallel]{SerialParam}} which uses a single core.
#'
#' @details
-#' This function is just a wrapper around \code{\link[scDD]{preprocess}} that
-#' takes the output and converts it to a SCDDParams object. See
-#' \code{\link[scDD]{preprocess}} for details.
+#' This function applies \code{\link[scDD]{preprocess}} to the counts then uses
+#' \code{\link[scDD]{scDD}} to estimate the numbers of each gene type to
+#' simulate. The output is then converted to a SCDDParams object. See
+#' \code{\link[scDD]{preprocess}} and \code{\link[scDD]{scDD}} for details.
#'
#' @return SCDDParams object containing the estimated parameters.
#'
#' @examples
+#' \dontrun{
#' data("sc_example_counts")
#' conditions <- sample(1:2, ncol(sc_example_counts), replace = TRUE)
#' params <- scDDEstimate(sc_example_counts, conditions)
#' params
+#' }
+#' @importFrom BiocParallel SerialParam
#' @export
-scDDEstimate <- function(counts, conditions, params = newSCDDParams()) {
+scDDEstimate <- function(counts, conditions, params = newSCDDParams(),
+ BPPARAM = SerialParam()) {
UseMethod("scDDEstimate")
}
#' @rdname scDDEstimate
#' @export
-scDDEstimate.SCESet <- function(counts, conditions, params = newSCDDParams()) {
+scDDEstimate.SCESet <- function(counts, conditions, params = newSCDDParams(),
+ BPPARAM = SerialParam()) {
counts <- scater::counts(counts)
scDDEstimate(counts, conditions, params)
}
@@ -35,13 +44,17 @@ scDDEstimate.SCESet <- function(counts, conditions, params = newSCDDParams()) {
#' @rdname scDDEstimate
#' @importFrom methods as
#' @export
-scDDEstimate.matrix <- function(counts, conditions, params = newSCDDParams()) {
+scDDEstimate.matrix <- function(counts, conditions, params = newSCDDParams(),
+ BPPARAM = SerialParam()) {
if (!requireNamespace("scDD", quietly = TRUE)) {
stop("The scDD simulation requires the 'scDD' package.")
}
checkmate::assertClass(params, "SCDDParams")
+ checkmate::assertMatrix(counts, mode = "numeric", any.missing = FALSE,
+ min.rows = 1, min.cols = 1, row.names = "unique",
+ col.names = "unique")
checkmate::assertIntegerish(conditions, len = ncol(counts), lower = 1,
upper = 2)
@@ -59,8 +72,29 @@ scDDEstimate.matrix <- function(counts, conditions, params = newSCDDParams()) {
SCdat <- SummarizedExperiment::SummarizedExperiment(assays = assays,
colData = colData)
- params <- setParams(params, nCells = round(dim(SCdat)[2] / 2),
- SCdat = SCdat)
+ SCdat <- scDD::scDD(SCdat, testZeroes = FALSE, param = BPPARAM)
+
+ res <- scDD::results(SCdat)
+ res <- res[!is.na(res$DDcategory), ]
+ dd.cats <- table(res$DDcategory)
+
+ not.dd <- res$DDcategory == "NS"
+ nDE <- ifelse("DE" %in% names(dd.cats), dd.cats["DE"], 0)
+ nDP <- ifelse("DP" %in% names(dd.cats), dd.cats["DP"], 0)
+ nDM <- ifelse("DM" %in% names(dd.cats), dd.cats["DM"], 0)
+ nDB <- ifelse("DB" %in% names(dd.cats), dd.cats["DB"], 0)
+ nEP <- sum(res$Clusters.c1[not.dd] > 1 & res$Clusters.c2[not.dd] > 1)
+ nEE <- nrow(counts) - nDE - nDP - nDM - nDB - nEP
+
+ params <- setParams(params,
+ nCells = round(dim(SCdat)[2] / 2),
+ SCdat = SCdat,
+ nDE = nDE,
+ nDP = nDP,
+ nDM = nDM,
+ nDB = nDB,
+ nEE = nEE,
+ nEP = nEP)
return(params)
-}
\ No newline at end of file
+}
diff --git a/man/SCDDParams.Rd b/man/SCDDParams.Rd
index 8e3b2ea..aef9df9 100644
--- a/man/SCDDParams.Rd
+++ b/man/SCDDParams.Rd
@@ -14,24 +14,24 @@ S4 class that holds parameters for the scDD simulation.
The SCDD simulation uses the following parameters:
\describe{
- \item{\code{[nGenes]}}{The number of genes to simulate (not used).}
+ \item{\code{nGenes}}{The number of genes to simulate (not used).}
\item{\code{nCells}}{The number of cells to simulate in each condition.}
\item{\code{[seed]}}{Seed to use for generating random numbers.}
\item{\code{SCdat}}{
\code{\link[SummarizedExperiment]{SummarizedExperiment}} containing real
data.}
- \item{\code{[nDE]}}{Number of DE genes to simulate.}
- \item{\code{[nDP]}}{Number of DP genes to simulate.}
- \item{\code{[nDM]}}{Number of DM genes to simulate.}
- \item{\code{[nDB]}}{Number of DB genes to simulate.}
- \item{\code{[nEE]}}{Number of EE genes to simulate.}
- \item{\code{[nEP]}}{Number of EP genes to simulate.}
+ \item{\code{nDE}}{Number of DE genes to simulate.}
+ \item{\code{nDP}}{Number of DP genes to simulate.}
+ \item{\code{nDM}}{Number of DM genes to simulate.}
+ \item{\code{nDB}}{Number of DB genes to simulate.}
+ \item{\code{nEE}}{Number of EE genes to simulate.}
+ \item{\code{nEP}}{Number of EP genes to simulate.}
\item{\code{[sd.range]}}{Interval for fold change standard deviations.}
\item{\code{[modeFC]}}{Values for DP, DM and DB mode fold changes.}
\item{\code{[varInflation]}}{Variance inflation factors for each
- condition. If all equal to 1 will be set to \code{NULL} (default)}
+ condition. If all equal to 1 will be set to \code{NULL} (default).}
\item{\code{[condition]}}{String giving the column that represents
- biological group of interest}
+ biological group of interest.}
}
The parameters not shown in brackets can be estimated from real data using
diff --git a/man/scDDEstimate.Rd b/man/scDDEstimate.Rd
index c0e3094..259a698 100644
--- a/man/scDDEstimate.Rd
+++ b/man/scDDEstimate.Rd
@@ -6,11 +6,14 @@
\alias{scDDEstimate.matrix}
\title{Estimate scDD simulation parameters}
\usage{
-scDDEstimate(counts, conditions, params = newSCDDParams())
+scDDEstimate(counts, conditions, params = newSCDDParams(),
+ BPPARAM = SerialParam())
-\method{scDDEstimate}{SCESet}(counts, conditions, params = newSCDDParams())
+\method{scDDEstimate}{SCESet}(counts, conditions, params = newSCDDParams(),
+ BPPARAM = SerialParam())
-\method{scDDEstimate}{matrix}(counts, conditions, params = newSCDDParams())
+\method{scDDEstimate}{matrix}(counts, conditions, params = newSCDDParams(),
+ BPPARAM = SerialParam())
}
\arguments{
\item{counts}{either a counts matrix or an SCESet object containing count
@@ -20,6 +23,10 @@ data to estimate parameters from.}
Conditions can be 1 or 2.}
\item{params}{SCDDParams object to store estimated values in.}
+
+\item{BPPARAM}{A \code{\link[BiocParallel]{BiocParallelParam}} instance
+giving the parallel back-end to be used. Default is
+\code{\link[BiocParallel]{SerialParam}} which uses a single core.}
}
\value{
SCDDParams object containing the estimated parameters.
@@ -28,13 +35,16 @@ SCDDParams object containing the estimated parameters.
Estimate simulation parameters for the scDD simulation from a real dataset.
}
\details{
-This function is just a wrapper around \code{\link[scDD]{preprocess}} that
-takes the output and converts it to a SCDDParams object. See
-\code{\link[scDD]{preprocess}} for details.
+This function applies \code{\link[scDD]{preprocess}} to the counts then uses
+\code{\link[scDD]{scDD}} to estimate the numbers of each gene type to
+simulate. The output is then converted to a SCDDParams object. See
+\code{\link[scDD]{preprocess}} and \code{\link[scDD]{scDD}} for details.
}
\examples{
+\dontrun{
data("sc_example_counts")
conditions <- sample(1:2, ncol(sc_example_counts), replace = TRUE)
params <- scDDEstimate(sc_example_counts, conditions)
params
}
+}
diff --git a/man/summariseDiff.Rd b/man/summariseDiff.Rd
index 3131547..cd6cb62 100644
--- a/man/summariseDiff.Rd
+++ b/man/summariseDiff.Rd
@@ -10,18 +10,17 @@ summariseDiff(diff)
\item{diff}{Output from \code{\link{diffSCESets}}}
}
\value{
-List with MADs, ranks and both combined in long format
+data.frame with MADs, MAEs, RMSEs, scaled statistics and ranks
}
\description{
-Summarise the results of \code{\link{diffSCESets}}. The various
-properties are sorted, differences calculated, the Median Absolute Deviation
-taken as the summary statistic and the ranks calculated.
+Summarise the results of \code{\link{diffSCESets}}. Calculates the Median
+Absolute Deviation (MAD), Mean Absolute Error (MAE) and Root Mean Squared
+Error (RMSE) for the various properties and ranks them.
}
\examples{
sim1 <- splatSimulate(nGenes = 1000, batchCells = 20)
sim2 <- simpleSimulate(nGenes = 1000, nCells = 20)
difference <- diffSCESets(list(Splat = sim1, Simple = sim2), ref = "Simple")
summary <- summariseDiff(difference)
-names(summary)
-head(summary$Long)
+head(summary)
}
--
GitLab