Package 'CSOA' reference manual

Title:	Calculate per-cell gene signature scores in scRNA-seq data using cell set overlaps
Description:	Cell Set Overlap Analysis (CSOA) is a tool for calculating per-cell gene signature scores in an scRNA-seq dataset. CSOA constructs a set for each gene in the signature, consisting of the cells that highly express the gene. Next, all overlaps of pairs of cell sets are computed, ranked, filtered and scored. The CSOA per-cell score is calculated by summing up all products of the overlap scores and the min-max-normalized expression of the two involved genes. CSOA can run on a Seurat object, a SingleCellExperiment object, a matrix and a dgCMatrix.
Authors:	Andrei-Florian Stoica [aut, cre] (ORCID: <https://orcid.org/0000-0002-5253-0826>)
Maintainer:	Andrei-Florian Stoica <[email protected]>
License:	MIT + file LICENSE
Version:	1.3.1
Built:	2026-05-06 11:09:39 UTC
Source:	https://github.com/bioc/CSOA

Attach CSOA scores to object

Description

This function attaches the data frame of CSOA scores to the input object.

Usage

## Default S3 method:
attachCellScores(scObj, scoreDF, ...)

## S3 method for class 'Seurat'
attachCellScores(scObj, scoreDF, ...)

## S3 method for class 'SingleCellExperiment'
attachCellScores(scObj, scoreDF, ...)

## S3 method for class 'matrix'
attachCellScores(scObj, scoreDF, ...)

## S3 method for class 'dgCMatrix'
attachCellScores(scObj, scoreDF, ...)

attachCellScores(scObj, ...)
## Default S3 method:
attachCellScores(scObj, scoreDF, ...)

## S3 method for class 'Seurat'
attachCellScores(scObj, scoreDF, ...)

## S3 method for class 'SingleCellExperiment'
attachCellScores(scObj, scoreDF, ...)

## S3 method for class 'matrix'
attachCellScores(scObj, scoreDF, ...)

## S3 method for class 'dgCMatrix'
attachCellScores(scObj, scoreDF, ...)

attachCellScores(scObj, ...)

Arguments

scObj

A Seurat object, SingleCellExperiment object, or expression matrix.

scoreDF

Data frame of CSOA scores.

...

Additional arguments.

Details

If the input object is of the Seurat or SingleCellExpression class, it will be returned with added CSOA scores. Otherwise, a list containing the expression matrix and the CSOA scores data frame will be returned.

Value

A Seurat object with CSOA scores added to metadata.

A SingleCellExperiment object with CSOA scores added to colData.

A list containing the expression matrix and the CSOA scores data frame.

Examples

library(Seurat)
mat <- matrix(0, 500, 300)
rownames(mat) <- paste0('G', seq(500))
colnames(mat) <- paste0('C', seq(300))
mat[sample(8000)] <- sample(20, 8000, TRUE)
seuratObj <- CreateSeuratObject(mat)
seuratObj <- NormalizeData(seuratObj)
scores <- data.frame(CSOA = runif(300))
seuratObj <- attachCellScores(seuratObj, scores)
head(seuratObj$CSOA)

library(Seurat)
mat <- matrix(0, 500, 300)
rownames(mat) <- paste0('G', seq(500))
colnames(mat) <- paste0('C', seq(300))
mat[sample(8000)] <- sample(20, 8000, TRUE)
seuratObj <- CreateSeuratObject(mat)
seuratObj <- NormalizeData(seuratObj)
scores <- data.frame(CSOA = runif(300))
seuratObj <- attachCellScores(seuratObj, scores)
head(seuratObj$CSOA)

Plot a simple heatmap

Description

This function plots a simple heatmap, with clustering but no dendograms.

Usage

basicHeatmap(
  mat,
  aesNames = c("x", "y", "Score"),
  title = "Heatmap",
  axisTextSize = 7,
  palette = paletteer_c("grDevices::Plasma", 30),
  ...
)
basicHeatmap(
  mat,
  aesNames = c("x", "y", "Score"),
  title = "Heatmap",
  axisTextSize = 7,
  palette = paletteer_c("grDevices::Plasma", 30),
  ...
)

Arguments

mat

A matrix.

aesNames

A character vector of length 3 representing the y, x and fill aes elements.

title

Plot title.

axisTextSize

Axis text size.

palette

Color palette.

...

Other arguments passed to henna::centerTitle.

Value

A ggplot object.

Examples

mat <- matrix(0, 10, 20)
mat[sample(length(mat), 50)] <- runif(50, max = 2.5)
basicHeatmap(mat)

mat <- matrix(0, 10, 20)
mat[sample(length(mat), 50)] <- runif(50, max = 2.5)
basicHeatmap(mat)

Remove overlap pairs with low Jaccard scores

Description

This function iteratively removes all overlap pairs with neighbor Jaccard score below a fixed cutoff until no overlap pairs can be removed. Subsequently, overlap ranks are recalculated.

Usage

breakWeakTies(
  overlapDF,
  cutoff = 1/3,
  doConnComp = FALSE,
  mtMethod = c("BY", "BH")
)
breakWeakTies(
  overlapDF,
  cutoff = 1/3,
  doConnComp = FALSE,
  mtMethod = c("BY", "BH")
)

Arguments

overlapDF

An overlap data frame.

cutoff

A cutoff used in the filtering of edges with low Jaccard scores.

doConnComp

Whether to calculate the connected components.

mtMethod

Multiple testing correction method. Choose between Benjamini-Yekutieli ('BY') and Benjamini-Hochberg('BH'). Default is 'BY'.

Details

The functions removes overlaps for which the two involved genes record too few shared neighbors—genes whose cell set significantly overlaps with those of both overlap genes.

Value

An overlap data frame in which edges with low Jaccard scores have been removed.

Examples

overlapDF <- data.frame(gene1=paste0('G', c(1, 3, 7, 6, 8, 2, 4, 3, 4, 5)),
gene2=paste0('G', c(2, 7, 2, 5, 4, 5, 1, 2, 2, 8)),
ratio=runif(10, 2, 10),
pval=runif(10, 0, 1e-10))
breakWeakTies(overlapDF, cutoff=0.1)

overlapDF <- data.frame(gene1=paste0('G', c(1, 3, 7, 6, 8, 2, 4, 3, 4, 5)),
gene2=paste0('G', c(2, 7, 2, 5, 4, 5, 1, 2, 2, 8)),
ratio=runif(10, 2, 10),
pval=runif(10, 0, 1e-10))
breakWeakTies(overlapDF, cutoff=0.1)

Show the distribution of cell sets among cells

Description

This function returns a logical matrix that shows the representation of cell sets among all cells.

Usage

cellDistribution(cellSets, allCells)
cellDistribution(cellSets, allCells)

Arguments

cellSets

A list of character vectors.

allCells

Names of all cells in the dataset.

Value

A logical matrix with genes as rows and cells as columns.

Examples

cellSets <- list(c('A', 'H', 'J'),
c('B', 'D', 'E', 'F', 'J'),
c('C', 'I', 'L'))
allCells <- LETTERS[seq(15)]
cellDistribution(cellSets, allCells)

cellSets <- list(c('A', 'H', 'J'),
c('B', 'D', 'E', 'F', 'J'),
c('C', 'I', 'L'))
allCells <- LETTERS[seq(15)]
cellDistribution(cellSets, allCells)

Calculates the significance of overlaps of pairs of cells sets

Description

This function computes the statistical significance of overlaps of pairs of cell sets.

Usage

cellSetsOverlaps(cellSets, nCells, pairs = NULL, overlapFileName = NULL)
cellSetsOverlaps(cellSets, nCells, pairs = NULL, overlapFileName = NULL)

Arguments

cellSets

A list of character arrays.

nCells

The total number of cells in the Seurat object.

pairs

Pairs of cell sets to be assessed. If NULL (as default), all pairs will be assessed.

overlapFileName

The name of the file where the overlap data frame will be saved. This option can be used to save time when performing exploratory analyses such as trying different jaccardCutoff parameters in breakWeakTies. Default is NULL (the overlap data frame will not be saved).

Value

A data frame listing statistics for all cell set overlaps: cell set sizes, recorded and expected shared cells, the recorded-over-expected ratio and the hypergeometric p-value.

Examples

cellSets <- list(G1 = c('A', 'H', 'J'),
G2 = c('B', 'D', 'E', 'F', 'J'),
G3 = c('C', 'I', 'L'))
cellSetsOverlaps(cellSets, 40)

cellSets <- list(G1 = c('A', 'H', 'J'),
G2 = c('B', 'D', 'E', 'F', 'J'),
G3 = c('C', 'I', 'L'))
cellSetsOverlaps(cellSets, 40)

Extracts the data expression matrix from object

Description

This function extracts the data expression matrix from object as a non-sparse matrix. Selected genes can be specified as input.

Usage

expMat(scObj, ...)

## Default S3 method:
expMat(scObj, genes = NULL, ...)

## S3 method for class 'Seurat'
expMat(scObj, ...)

## S3 method for class 'SingleCellExperiment'
expMat(scObj, ...)

## S3 method for class 'dgCMatrix'
expMat(scObj, ...)

## S3 method for class 'matrix'
expMat(scObj, ...)
expMat(scObj, ...)

## Default S3 method:
expMat(scObj, genes = NULL, ...)

## S3 method for class 'Seurat'
expMat(scObj, ...)

## S3 method for class 'SingleCellExperiment'
expMat(scObj, ...)

## S3 method for class 'dgCMatrix'
expMat(scObj, ...)

## S3 method for class 'matrix'
expMat(scObj, ...)

Arguments

scObj

A Seurat object, SingleCellExperiment object, or expression matrix.

...

Additional arguments.

genes

Genes retained in the expression matrix. If NULL, all genes will be retained

Value

An expression matrix.

Examples

library(Seurat)
mat <- matrix(0, 6, 4)
mat[sample(length(mat), 7)] <- sample(3, 7, TRUE)
seuratObj <- CreateSeuratObject(counts = mat)
seuratObj <- NormalizeData(seuratObj)
expMat(seuratObj)

library(Seurat)
mat <- matrix(0, 6, 4)
mat[sample(length(mat), 7)] <- sample(3, 7, TRUE)
seuratObj <- CreateSeuratObject(counts = mat)
seuratObj <- NormalizeData(seuratObj)
expMat(seuratObj)

A feature plot with a more distinctive color scheme.

Description

This function customizes the appearance of Seurat::FeaturePlot for improved distinctiveness and aesthetics.

Usage

featureWes(
  seuratObj,
  feature,
  title = feature,
  idClass = NULL,
  labelSize = 3.5,
  titleSize = 12,
  palette = paletteer_d("wesanderson::Royal1")[c(3, 2)],
  ...
)
featureWes(
  seuratObj,
  feature,
  title = feature,
  idClass = NULL,
  labelSize = 3.5,
  titleSize = 12,
  palette = paletteer_d("wesanderson::Royal1")[c(3, 2)],
  ...
)

Arguments

seuratObj

A Seurat object.

feature

Seurat feature.

title

Plot title.

idClass

Column to be used for labelling. If NULL, no column-based labels will be generated.

labelSize

Size of labels. Ignored if idClass is NULL.

titleSize

Title size.

palette

Color palette.

...

Additional arguments passed to Seurat::FeaturePlot.

Value

A ggplot object.

Examples

library(Seurat)
mat <- matrix(0, 3000, 800)
mat[sample(length(mat), 90000)] <- sample(8, 90000, TRUE)
seuratObj <- CreateSeuratObject(counts = mat)
seuratObj <- FindVariableFeatures(seuratObj, nfeatures=200)
seuratObj <- NormalizeData(seuratObj)
seuratObj <- ScaleData(seuratObj)
seuratObj <- RunPCA(seuratObj, verbose=FALSE)
seuratObj <- RunUMAP(seuratObj, dims=1:20, verbose=FALSE)
featureWes(seuratObj, 'Feature3')

library(Seurat)
mat <- matrix(0, 3000, 800)
mat[sample(length(mat), 90000)] <- sample(8, 90000, TRUE)
seuratObj <- CreateSeuratObject(counts = mat)
seuratObj <- FindVariableFeatures(seuratObj, nfeatures=200)
seuratObj <- NormalizeData(seuratObj)
seuratObj <- ScaleData(seuratObj)
seuratObj <- RunPCA(seuratObj, verbose=FALSE)
seuratObj <- RunUMAP(seuratObj, dims=1:20, verbose=FALSE)
featureWes(seuratObj, 'Feature3')

Radial plot for an overlap data frame

Description

This function draws a radial plot for an overlap data frame to illustrate gene participation in top overlaps.

Usage

geneRadialPlot(
  overlapObj,
  title = "Top overlap genes plot",
  degreeLegendTitle = "Number of top overlaps",
  groupLegendTitle = "Group",
  extraCircles = 2,
  groupNames = NULL,
  cutoff = NULL,
  ...
)
geneRadialPlot(
  overlapObj,
  title = "Top overlap genes plot",
  degreeLegendTitle = "Number of top overlaps",
  groupLegendTitle = "Group",
  extraCircles = 2,
  groupNames = NULL,
  cutoff = NULL,
  ...
)

Arguments

overlapObj

An overlap data frame or list of overlap data frames.

title

Plot title.

degreeLegendTitle

The title of the degree legend.

groupLegendTitle

The title of the group legend. If NULL, no groups will be distinguished.

extraCircles

Number of extra circles to be displayed on the plot.

groupNames

Names of groups. If provided, must be a vector of the same length as the list of overlap data frames.

cutoff

Number of retained edges from each overlap data frame after refiltering. If NULL (as default), no refiltering will be performed.

...

Additional parameters passed to henna::radialPlot.

Details

The function can separate genes by groups. The groups can be, for instance, different gene sets, or different connected components of the same overlap data frame. A wrapper around henna::radialPlot

Value

A ggplot object.

Examples

edgesDF <- data.frame(gene1 = paste0('G', c(1, 2, 3, 4, 7, 8, 10,
11, 11, 10, 10, 10)),
gene2 = paste0('G', c(2, 5, 1, 8, 4, 9, 12,
13, 14, 13, 16, 14)))
edgesDF <- henna::connectedComponents(edgesDF, 'group')
geneRadialPlot(edgesDF, groupLegendTitle='Component', extraCircles=1)

edgesDF <- data.frame(gene1 = paste0('G', c(1, 2, 3, 4, 7, 8, 10,
11, 11, 10, 10, 10)),
gene2 = paste0('G', c(2, 5, 1, 8, 4, 9, 12,
13, 14, 13, 16, 14)))
edgesDF <- henna::connectedComponents(edgesDF, 'group')
geneRadialPlot(edgesDF, groupLegendTitle='Component', extraCircles=1)

Generate overlaps of cell sets for input genes

Description

This function constructs, for each gene in the expression matrix, a set of cells expressing the gene at or above the input percentile. Subsequently, overlaps of pairs of the constructed cell sets are assessed for statistical significance.

Usage

generateOverlaps(
  geneSetExp,
  percentile = 90,
  pairs = NULL,
  overlapFileName = NULL
)
generateOverlaps(
  geneSetExp,
  percentile = 90,
  pairs = NULL,
  overlapFileName = NULL
)

Arguments

geneSetExp

A gene expression non-sparse matrix with the rows restricted to the genes for which cell sets will be computed.

percentile

A positive number under 100.

pairs

Pairs of cell sets to be assessed. If NULL (as default), all pairs will be assessed.

overlapFileName

Details

Wrapper around percentileSets and cellSetsOverlaps.

Value

A data frame listing statistics for all cell set overlaps

Examples

mat <- matrix(0, 2000, 500)
rownames(mat) <- paste0('G', seq(2000))
colnames(mat) <- paste0('C', seq(500))
mat[sample(length(mat), 270000)] <- sample(50, 270000, TRUE)
mat <- mat[paste0('G', sample(2000, 5)), ]
generateOverlaps(mat)

mat <- matrix(0, 2000, 500)
rownames(mat) <- paste0('G', seq(2000))
colnames(mat) <- paste0('C', seq(500))
mat[sample(length(mat), 270000)] <- sample(50, 270000, TRUE)
mat <- mat[paste0('G', sample(2000, 5)), ]
generateOverlaps(mat)

Get all unordered pairs of two elements from a vector

Description

This function returns all unorderded pairs of two elements from a vector.

Usage

getPairs(v)
getPairs(v)

Arguments

v

A vector.

Value

A list of vectors of length 2.

Examples

v <- c('ASD', 'VBN', 'HJKL')
getPairs(v)

v <- c('ASD', 'VBN', 'HJKL')
getPairs(v)

Plot the selection of overlaps

Description

This function illustrates the process of selecting the overlap rank cutoff by plotting rank frequencies against ranks and showcasing the convex hull of the rank-frequency points.

Usage

overlapCutoffPlot(
  overlapDF,
  title = "Overlap cutoff plot",
  palette = c("purple", "yellow"),
  hullWidth = 0.8,
  xLab = "Overlap rank",
  yLab = "Frequency",
  legendLabs = c("Accepted overlaps", "Discarded overlaps"),
  pointShape = 24,
  ...
)
overlapCutoffPlot(
  overlapDF,
  title = "Overlap cutoff plot",
  palette = c("purple", "yellow"),
  hullWidth = 0.8,
  xLab = "Overlap rank",
  yLab = "Frequency",
  legendLabs = c("Accepted overlaps", "Discarded overlaps"),
  pointShape = 24,
  ...
)

Arguments

overlapDF

Processed overlap data frame created with processOverlaps.

title

Plot title.

palette

Color palette. Must have two colors, the first one representing accepted overlaps and the other representing discarded overlaps.

hullWidth

Width of the convex hull.

xLab

x axis label.

yLab

y axis label.

legendLabs

Legend labels.

pointShape

Point shape.

...

Additional arguments passed to henna::hullPlot.

Details

A wrapper around henna::hullPlot.

Value

A ggplot object.

Examples

overlapDF <- data.frame(gene1=paste0('G', c(1, 3, 7, 6, 8, 2, 4, 3, 4, 5)),
gene2=paste0('G', c(2, 7, 2, 5, 4, 5, 1, 2, 2, 8)),
rank=c(1, 2, 3, 4, 4, 6, 7, 7, 7, 10))
overlapCutoffPlot(overlapDF)

overlapDF <- data.frame(gene1=paste0('G', c(1, 3, 7, 6, 8, 2, 4, 3, 4, 5)),
gene2=paste0('G', c(2, 7, 2, 5, 4, 5, 1, 2, 2, 8)),
rank=c(1, 2, 3, 4, 4, 6, 7, 7, 7, 10))
overlapCutoffPlot(overlapDF)

Get all genes from an overlap data frame

Description

This function gets all genes from an overlap data frame.

Usage

overlapGenes(overlapDF, components = NULL)
overlapGenes(overlapDF, components = NULL)

Arguments

overlapDF

Overlap data frame.

components

A numeric vector representing the connected components of the overlap data frame graph.

Value

A character vector of genes.

Examples

overlapDF <- data.frame(gene1 = paste0('G', c(1, 2, 3)),
gene1 = paste0('G', c(2, 7, 8)))
overlapGenes(overlapDF)

overlapDF <- data.frame(gene1 = paste0('G', c(1, 2, 3)),
gene1 = paste0('G', c(2, 7, 8)))
overlapGenes(overlapDF)

Plot the overlaps as a network

Description

This function plots the graph of the overlap data frame, with genes as vertices and overlaps as edges.

Usage

overlapNetworkPlot(
  overlapDF,
  title = "Top overlaps network plot",
  nodeColor = "orange",
  edgeColor = "green4",
  ...
)
overlapNetworkPlot(
  overlapDF,
  title = "Top overlaps network plot",
  nodeColor = "orange",
  edgeColor = "green4",
  ...
)

Arguments

overlapDF

Overlap data frame.

title

Plot title.

nodeColor

The color of nodes. If NULL, the default henna::networkPlot color scheme will be used, which uses different colors for nodes belonging to different connected components.

edgeColor

The color of edges.

...

Additional parameters passed to henna::networkPlot.

Details

A thin wrapper around henna::networkPlot.

Value

An overlap network plot.

Examples

overlapDF <- data.frame(gene1 = paste0('G', c(1, 2, 5, 6, 7, 17)),
gene2 = paste0('G', c(2, 5, 8, 11, 11, 11)),
rank = c(1, 1, 3, 3, 3, 3))
overlapNetworkPlot(overlapDF)

overlapDF <- data.frame(gene1 = paste0('G', c(1, 2, 5, 6, 7, 17)),
gene2 = paste0('G', c(2, 5, 8, 11, 11, 11)),
rank = c(1, 1, 3, 3, 3, 3))
overlapNetworkPlot(overlapDF)

Extract gene pairs from overlap data frame

Description

This function extracts the gene pairs from an overlap data frame.

Usage

overlapPairs(overlapDF)
overlapPairs(overlapDF)

Arguments

overlapDF

Overlap data frame.

Value

A list of gene pairs.

Examples

overlapDF <- data.frame(gene1 = paste0('G', c(1, 2, 3)),
gene1 = paste0('G', c(2, 7, 8)))
overlapPairs(overlapDF)

overlapDF <- data.frame(gene1 = paste0('G', c(1, 2, 3)),
gene1 = paste0('G', c(2, 7, 8)))
overlapPairs(overlapDF)

Generates cell expressing input genes at an input percentile

Description

This function constructs, for each gene in the expression matrix, a set of cells expressing the gene at or above the input percentile.

Usage

percentileSets(geneSetExp, percentile = 90)
percentileSets(geneSetExp, percentile = 90)

Arguments

geneSetExp

A gene expression non-sparse matrix with the rows restricted to the genes for which cell sets will be computed.

percentile

A positive number under 100.

Value

A named list of character vectors of length equaling the number of input genes. Each vector stores the cells expressing the gene at or above the input percentile.

Examples

mat <- matrix(0, 1000, 500)
rownames(mat) <- paste0('G', seq(1000))
colnames(mat) <- paste0('C', seq(500))
mat[sample(length(mat), 70000)] <- sample(50, 70000, TRUE)
mat <- mat[paste0('G', sample(1000, 3)), ]
percentileSets(mat)

mat <- matrix(0, 1000, 500)
rownames(mat) <- paste0('G', seq(1000))
colnames(mat) <- paste0('C', seq(500))
mat[sample(length(mat), 70000)] <- sample(50, 70000, TRUE)
mat <- mat[paste0('G', sample(1000, 3)), ]
percentileSets(mat)

Process data frame of overlaps of cell sets

Description

This function filters, ranks and scores previously generated overlaps of cell sets.

Usage

processOverlaps(
  overlapDF,
  mtMethod = c("BY", "BH"),
  jaccardCutoff = NULL,
  osMethod = c("log", "minmax"),
  ...
)
processOverlaps(
  overlapDF,
  mtMethod = c("BY", "BH"),
  jaccardCutoff = NULL,
  osMethod = c("log", "minmax"),
  ...
)

Arguments

overlapDF

Overlap data frame.

mtMethod

Multiple testing correction method. Choose between Benjamini-Yekutieli ('BY') and Benjamini-Hochberg('BH'). Default is 'BY'.

jaccardCutoff

A cutoff used in the filtering of edges with low Jaccard scores. If NULL (as default), no filtering of such edges will be performed.

osMethod

Method used to compute overlap scores. Options are "log" and "minmax".

...

Additional arguments passed to mtCorrectDF.

Details

Wrapper around byCorrectDF, rankOverlaps, prepareFiltering, filterOverlaps and scoreOverlaps.

If jaccardCutoff is not NULL, it also calls breakWeakTies between filterOverlaps and scoreOverlaps.

Value

A data frame consisting of filtered, ranked and scored cell sets overlaps

Examples

overlapDF <- data.frame(gene1=paste0('G',
c(1, 3, 7, 6, 8, 2, 4, 3, 4, 5)),
gene2=paste0('G',
c(2, 7, 2, 5, 4, 5, 1, 2, 2, 8)),
ratio=runif(10, 2, 10),
pval=runif(10, 0, 1e-10))
processOverlaps(overlapDF)

overlapDF <- data.frame(gene1=paste0('G',
c(1, 3, 7, 6, 8, 2, 4, 3, 4, 5)),
gene2=paste0('G',
c(2, 7, 2, 5, 4, 5, 1, 2, 2, 8)),
ratio=runif(10, 2, 10),
pval=runif(10, 0, 1e-10))
processOverlaps(overlapDF)

Read and delete a .qs2 file

Description

This functions reads a .qs2 file, deletes it, and returns its content.

Usage

qGrab(qs2File)
qGrab(qs2File)

Arguments

qs2File

Name of .qs2 file with path.

Value

The content of the .qs2 file.

Examples

library(qs2)
qs_save(c(1, 2, 3), 'temp.qs2')
qGrab('temp.qs2')

library(qs2)
qs_save(c(1, 2, 3), 'temp.qs2')
qGrab('temp.qs2')

Run the CSOA pipeline

Description

This function generates cell set overlaps for input gene sets based on percentiles of gene expression, computes the significance of these overlaps, ranks, filters and scores the overlaps, and builds a per-cell score by summing the products of overlap scores and the min-max-normalized expression of the corresponding pairs of genes.

Usage

runCSOA(
  scObj,
  geneSets,
  percentile = 90,
  mtMethod = c("BY", "BH"),
  jaccardCutoff = NULL,
  osMethod = c("log", "minmax"),
  overlapFileName = NULL,
  pairFileTemplate = NULL,
  keepOverlapOrder = FALSE,
  ...
)
runCSOA(
  scObj,
  geneSets,
  percentile = 90,
  mtMethod = c("BY", "BH"),
  jaccardCutoff = NULL,
  osMethod = c("log", "minmax"),
  overlapFileName = NULL,
  pairFileTemplate = NULL,
  keepOverlapOrder = FALSE,
  ...
)

Arguments

scObj

A Seurat object, SingleCellExperiment object, or expression matrix.

geneSets

Named list of character vectors of which each must contain at least two genes.

percentile

A positive number under 100.

mtMethod

Multiple testing correction method. Choose between Benjamini-Yekutieli ('BY') and Benjamini-Hochberg('BH'). Default is 'BY'.

jaccardCutoff

A cutoff used in the filtering of edges with low Jaccard scores. If NULL (as default), no filtering of such edges will be performed.

osMethod

Method used to compute overlap scores. Options are "log" and "minmax".

overlapFileName

pairFileTemplate

Character object used in the naming of the files where the pair data frames will be saved. Default is NULL (the pair data frames will not be saved).

keepOverlapOrder

Keep the rank-based order of overlaps in the pair score file, as opposed to changing it to a pair score-based order. Ignored if pairFileTemplate is NULL.

...

Additional arguments.

Details

Wrapper around expMat, generateOverlaps, scoreCells and attachCellScores.

Value

An object of the same class as scObj with per-gene-set CSOA scores assigned for each cell.

Examples

mat <- matrix(0, 500, 300)
rownames(mat) <- paste0('G', seq(500))
colnames(mat) <- paste0('C', seq(300))
mat[sample(8000)] <- runif(8000, max=13)
genes <- paste0('G', seq(200))
mat[genes, 20:50] <- matrix(runif(200 * 31, min = 14, max = 15),
nrow = 200, ncol = 31)
geneSet1 <- paste0('G', seq(1, 150))
geneSet2 <- paste0('G', seq(50, 200))
df <- runCSOA(mat, list(a = geneSet1, b = geneSet2))
head(df)

mat <- matrix(0, 500, 300)
rownames(mat) <- paste0('G', seq(500))
colnames(mat) <- paste0('C', seq(300))
mat[sample(8000)] <- runif(8000, max=13)
genes <- paste0('G', seq(200))
mat[genes, 20:50] <- matrix(runif(200 * 31, min = 14, max = 15),
nrow = 200, ncol = 31)
geneSet1 <- paste0('G', seq(1, 150))
geneSet2 <- paste0('G', seq(50, 200))
df <- runCSOA(mat, list(a = geneSet1, b = geneSet2))
head(df)

Generate CSOA scores from overlap data frame and list of pairs

Description

This function scores an overlap data frame using its associated list of pairs. The overlap data frame is split based on the overlaps corresponding to each gene set and scored, and the output is rejoined as a data frame.

Usage

scoreCells(
  geneSetExp,
  overlapDF,
  setPairs,
  geneSetNames,
  mtMethod = c("BY", "BH"),
  jaccardCutoff = NULL,
  osMethod = c("log", "minmax"),
  pairFileTemplate = NULL,
  keepOverlapOrder = FALSE,
  ...
)
scoreCells(
  geneSetExp,
  overlapDF,
  setPairs,
  geneSetNames,
  mtMethod = c("BY", "BH"),
  jaccardCutoff = NULL,
  osMethod = c("log", "minmax"),
  pairFileTemplate = NULL,
  keepOverlapOrder = FALSE,
  ...
)

Arguments

geneSetExp

A gene expression non-sparse matrix with the rows restricted to the genes for which cell sets will be computed.

overlapDF

Overlap data frame.

setPairs

A list of overlaps corresponding to each input gene set.

geneSetNames

Character vector of names of gene sets.

mtMethod

Multiple testing correction method. Choose between Benjamini-Yekutieli ('BY') and Benjamini-Hochberg('BH'). Default is 'BY'.

jaccardCutoff

A cutoff used in the filtering of edges with low Jaccard scores. If NULL (as default), no filtering of such edges will be performed.

osMethod

Method used to compute overlap scores. Options are "log" and "minmax".

pairFileTemplate

Character object used in the naming of the files where the pair data frames will be saved. Default is NULL (the pair data frames will not be saved).

keepOverlapOrder

Keep the rank-based order of overlaps in the pair score file, as opposed to changing it to a pair score-based order. Ignored if pairFileTemplate is NULL.

...

Additional arguments passed to mtCorrectDF.

Details

This function calls scoreCells to score each gene set data frame split from the full overlap data frame.

Value

A data frame whose columns correspond to the CSOA scores of the input gene sets.

Examples

mat <- matrix(0, 500, 300)
rownames(mat) <- paste0('G', seq(500))
colnames(mat) <- paste0('C', seq(300))
mat[sample(8000)] <- runif(8000, max=13)
genes <- paste0('G', seq(200))
mat[genes, 20:50] <- matrix(runif(200 * 31, min=14, max=15),
nrow=200, ncol=31)
geneSet1 <- paste0('G', seq(1, 150))
geneSet2 <- paste0('G', seq(50, 200))
geneSets <- list(geneSet1, geneSet2)
geneSets <- lapply(geneSets, sort)
setPairs <- lapply(geneSets, getPairs)
pairs <- Reduce(union, setPairs)
genes <- union(geneSet1, geneSet2)
mat <- mat[genes, ]
overlapDF <- generateOverlaps(mat, pairs=pairs)
scoreDF <- scoreCells(mat, overlapDF, setPairs, c('set1', 'set2'))
head(scoreDF)

mat <- matrix(0, 500, 300)
rownames(mat) <- paste0('G', seq(500))
colnames(mat) <- paste0('C', seq(300))
mat[sample(8000)] <- runif(8000, max=13)
genes <- paste0('G', seq(200))
mat[genes, 20:50] <- matrix(runif(200 * 31, min=14, max=15),
nrow=200, ncol=31)
geneSet1 <- paste0('G', seq(1, 150))
geneSet2 <- paste0('G', seq(50, 200))
geneSets <- list(geneSet1, geneSet2)
geneSets <- lapply(geneSets, sort)
setPairs <- lapply(geneSets, getPairs)
pairs <- Reduce(union, setPairs)
genes <- union(geneSet1, geneSet2)
mat <- mat[genes, ]
overlapDF <- generateOverlaps(mat, pairs=pairs)
scoreDF <- scoreCells(mat, overlapDF, setPairs, c('set1', 'set2'))
head(scoreDF)

Run CSOA separately on the connected components of the overlap graph

Description

This function runs CSOA on the connected components of the graph having the filtered overlaps as edges.

Usage

scoreModules(
  scObj,
  networkDF,
  components,
  colStrTemplate = "CSOA_component",
  ...
)
scoreModules(
  scObj,
  networkDF,
  components,
  colStrTemplate = "CSOA_component",
  ...
)

Arguments

scObj

A Seurat object, SingleCellExperiment object, or expression matrix.

networkDF

A data frame with gene1, gene2 and component columns.

components

A numeric vector representing the connected components of the overlap data frame graph.

colStrTemplate

Character used in the naming of the component gene sets.

...

Additional parameters passed to runCSOAMultiple.

Value

An object of the same class as scObj with CSOA scores corresponding to the genes defining each connected components assigned for each cell.

Examples

mat <- matrix(0, 500, 300)
rownames(mat) <- paste0('G', seq(500))
colnames(mat) <- paste0('C', seq(300))
mat[sample(8000)] <- runif(8000, max=13)
genes1 <- paste0('G', seq(100))
mat[genes1, 20:50] <- matrix(runif(100 * 31, min = 14, max = 15),
nrow = 100, ncol = 31)
genes2 <- paste0('G', seq(101, 200))
mat[genes2, 70:100] <- matrix(runif(100 * 31, min = 14, max = 15),
nrow = 100, ncol = 31)
genes <- union(genes1, genes2)
mat <- mat[genes, ]
overlapDF <- generateOverlaps(mat)
overlapDF <- processOverlaps(overlapDF)
overlapDF <- henna::connectedComponents(overlapDF)
df <- scoreModules(mat, overlapDF, unique(overlapDF$component))[[2]]
head(df)

mat <- matrix(0, 500, 300)
rownames(mat) <- paste0('G', seq(500))
colnames(mat) <- paste0('C', seq(300))
mat[sample(8000)] <- runif(8000, max=13)
genes1 <- paste0('G', seq(100))
mat[genes1, 20:50] <- matrix(runif(100 * 31, min = 14, max = 15),
nrow = 100, ncol = 31)
genes2 <- paste0('G', seq(101, 200))
mat[genes2, 70:100] <- matrix(runif(100 * 31, min = 14, max = 15),
nrow = 100, ncol = 31)
genes <- union(genes1, genes2)
mat <- mat[genes, ]
overlapDF <- generateOverlaps(mat)
overlapDF <- processOverlaps(overlapDF)
overlapDF <- henna::connectedComponents(overlapDF)
df <- scoreModules(mat, overlapDF, unique(overlapDF$component))[[2]]
head(df)

Package 'CSOA'

Help Index

Attach CSOA scores to object

Description

Usage

Arguments

Details

Value

Examples

Plot a simple heatmap

Description

Usage

Arguments

Value

Examples

Remove overlap pairs with low Jaccard scores

Description

Usage

Arguments

Details

Value

Examples

Show the distribution of cell sets among cells

Description

Usage

Arguments

Value

Examples

Calculates the significance of overlaps of pairs of cells sets

Description

Usage

Arguments

Value

Examples

Extracts the data expression matrix from object

Description

Usage

Arguments

Value

Examples

A feature plot with a more distinctive color scheme.

Description

Usage

Arguments

Value

Examples

Radial plot for an overlap data frame

Description

Usage

Arguments

Details

Value

Examples

Generate overlaps of cell sets for input genes

Description

Usage

Arguments

Details

Value

Examples

Get all unordered pairs of two elements from a vector

Description

Usage

Arguments

Value

Examples

Plot the selection of overlaps

Description

Usage

Arguments

Details

Value

Examples

Get all genes from an overlap data frame

Description

Usage

Arguments

Value

Examples

Plot the overlaps as a network