Correct unloading of the library

Description

Correct unloading of the library

Usage

.onUnload(libpath)

Arguments

Return the mode of a vector

Description

Return the mode of a vector

Usage

Mode(x)

Arguments

Value

the mode elements

Pagoda2 R6 class

Description

The class encompasses gene count matrices, providing methods for normalization, calculating embeddings, and differential expression.

Public fields

Methods

Public methods

• Pagoda2$new()

• Pagoda2$setCountMatrix()

• Pagoda2$adjustVariance()

• Pagoda2$makeKnnGraph()

• Pagoda2$getKnnClusters()

• Pagoda2$geneKnnbyPCA()

• Pagoda2$getHierarchicalDiffExpressionAspects()

• Pagoda2$makeGeneKnnGraph()

• Pagoda2$getDensityClusters()

• Pagoda2$getDifferentialGenes()

• Pagoda2$plotDiffGeneHeatmap()

• Pagoda2$getRefinedLibSizes()

• Pagoda2$plotGeneHeatmap()

• Pagoda2$plotEmbedding()

• Pagoda2$getOdGenes()

• Pagoda2$getNormalizedExpressionMatrix()

• Pagoda2$calculatePcaReduction()

• Pagoda2$expandOdGenes()

• Pagoda2$localPcaKnn()

• Pagoda2$testPathwayOverdispersion()

• Pagoda2$getEmbedding()

• Pagoda2$clone()

Method new()

Initialize Pagoda2 class

Usage

Pagoda2$new(
  x,
  modelType = "plain",
  n.cores = parallel::detectCores(logical = FALSE),
  verbose = TRUE,
  min.cells.per.gene = 0,
  trim = round(min.cells.per.gene/2),
  min.transcripts.per.cell = 10,
  batch = NULL,
  lib.sizes = NULL,
  log.scale = TRUE,
  keep.genes = NULL
)

Arguments

Returns

new Pagoda2 object

Examples

\donttest{ 
## Load pre-generated a dataset of 50 bone marrow cells as matrix
cm <- readRDS(system.file("extdata", "sample_BM1_50.rds", package="pagoda2"))
## Perform QC, i.e. filter any cells that
counts <- gene.vs.molecule.cell.filter(cm, min.cell.size=500)
rownames(counts) <- make.unique(rownames(counts))
## Generate Pagoda2 object 
p2_object <- Pagoda2$new(counts, log.scale=TRUE, min.cells.per.gene=10, n.cores=1) 
}

Method setCountMatrix()

Provide the initial count matrix, and estimate deviance residual matrix (correcting for depth and batch)

Usage

Pagoda2$setCountMatrix(
  countMatrix,
  depthScale = 1000,
  min.cells.per.gene = 0,
  trim = round(min.cells.per.gene/2),
  min.transcripts.per.cell = 10,
  lib.sizes = NULL,
  log.scale = FALSE,
  keep.genes = NULL,
  verbose = TRUE
)

Arguments

Returns

normalized count matrix (or if modelTye='raw', the unnormalized count matrix)

Method adjustVariance()

Adjust variance of the residual matrix, determine overdispersed sites This is done to normalize the extent to which genes with (very) different expression magnitudes will contribute to the downstream anlaysis.

Usage

Pagoda2$adjustVariance(
  gam.k = 5,
  alpha = 0.05,
  plot = FALSE,
  use.raw.variance = FALSE,
  use.unadjusted.pvals = FALSE,
  do.par = TRUE,
  max.adjusted.variance = 1000,
  min.adjusted.variance = 0.001,
  cells = NULL,
  verbose = TRUE,
  min.gene.cells = 0,
  persist = is.null(cells),
  n.cores = self$n.cores
)

Arguments

Returns

residual matrix with adjusted variance

Examples

\donttest{
## Load pre-generated a dataset of 50 bone marrow cells as matrix
cm <- readRDS(system.file("extdata", "sample_BM1_50.rds", package="pagoda2"))
## Perform QC, i.e. filter any cells that
counts <- gene.vs.molecule.cell.filter(cm, min.cell.size=500)
rownames(counts) <- make.unique(rownames(counts))
## Generate Pagoda2 object 
p2_object <- Pagoda2$new(counts,log.scale=TRUE, min.cells.per.gene=10, n.cores=1) 
## Normalize gene expression variance
p2_object$adjustVariance(plot=TRUE, gam.k=10)
}

Method makeKnnGraph()

Create k-nearest neighbor graph

Usage

Pagoda2$makeKnnGraph(
  k = 30,
  nrand = 1000,
  type = "counts",
  weight.type = "1m",
  odgenes = NULL,
  n.cores = self$n.cores,
  distance = "cosine",
  center = TRUE,
  x = NULL,
  p = NULL,
  var.scale = (type == "counts"),
  verbose = TRUE
)

Arguments

Returns

kNN graph, stored in self$graphs

Examples

\donttest{
## Load pre-generated a dataset of 50 bone marrow cells as matrix
cm <- readRDS(system.file("extdata", "sample_BM1_50.rds", package="pagoda2"))
## Perform QC, i.e. filter any cells that
counts <- gene.vs.molecule.cell.filter(cm, min.cell.size=300)
rownames(counts) <- make.unique(rownames(counts))
## Generate Pagoda2 object   
p2_object <- Pagoda2$new(counts,log.scale=TRUE, min.cells.per.gene=10, n.cores=1) 
## Normalize gene expression variance
p2_object$adjustVariance(plot=TRUE, gam.k=10)
## Generate a kNN graph of cells that will allow us to identify clusters of cells
p2_object$makeKnnGraph(k=20, center=FALSE, distance='L2')
}

Method getKnnClusters()

Calculate clusters based on the kNN graph

Usage

Pagoda2$getKnnClusters(
  type = "counts",
  method = igraph::multilevel.community,
  name = "community",
  n.cores = self$n.cores,
  g = NULL,
  min.cluster.size = 1,
  persist = TRUE,
  ...
)

Arguments

Returns

the community structure calculated from 'method'

Method geneKnnbyPCA()

Deprecated function. Use makeGeneKnnGraph() instead.

Usage

Pagoda2$geneKnnbyPCA()

Method getHierarchicalDiffExpressionAspects()

Take a given clustering and generate a hierarchical clustering

Usage

Pagoda2$getHierarchicalDiffExpressionAspects(
  type = "counts",
  groups = NULL,
  clusterName = NULL,
  method = "ward.D",
  dist = "pearson",
  persist = TRUE,
  z.threshold = 2,
  n.cores = self$n.cores,
  min.set.size = 5,
  verbose = TRUE
)

Arguments

Returns

hierarchical clustering

Method makeGeneKnnGraph()

Calculates gene Knn network for gene similarity

Usage

Pagoda2$makeGeneKnnGraph(
  nPcs = 100,
  center = TRUE,
  fastpath = TRUE,
  maxit = 1000,
  k = 30,
  n.cores = self$n.cores,
  verbose = TRUE
)

Arguments

Returns

graph with gene similarity

Method getDensityClusters()

Calculate density-based clusters

Usage

Pagoda2$getDensityClusters(
  type = "counts",
  embeddingType = NULL,
  name = "density",
  eps = 0.5,
  v = 0.7,
  s = 1,
  verbose = TRUE,
  ...
)

Arguments

Returns

density-based clusters

Method getDifferentialGenes()

Determine differentially expressed genes, comparing each group against all others using Wilcoxon rank sum test

Usage

Pagoda2$getDifferentialGenes(
  type = "counts",
  clusterType = NULL,
  groups = NULL,
  name = "customClustering",
  z.threshold = 3,
  upregulated.only = FALSE,
  verbose = FALSE,
  append.specificity.metrics = TRUE,
  append.auc = FALSE
)

Arguments

Returns

List with each element of the list corresponding to a cell group in the provided/used factor (i.e. factor levels) Each element of a list is a data frame listing the differentially epxressed genes (row names), with the following columns: Z - adjusted Z score, with positive values indicating higher expression in a given group compare to the rest M - log2 fold change highest- a boolean flag indicating whether the expression of a given gene in a given vcell group was on average higher than in every other cell group fe - fraction of cells in a given group having non-zero expression level of a given gene

Method plotDiffGeneHeatmap()

Plot heatmap of DE results

Usage

Pagoda2$plotDiffGeneHeatmap(
  type = "counts",
  clusterType = NULL,
  groups = NULL,
  n.genes = 100,
  z.score = 2,
  gradient.range.quantile = 0.95,
  inner.clustering = FALSE,
  gradientPalette = NULL,
  v = 0.8,
  s = 1,
  box = TRUE,
  drawGroupNames = FALSE,
  ...
)

Arguments

Returns

heatmap of DE results

Method getRefinedLibSizes()

Recalculate library sizes using robust regression within clusters

Usage

Pagoda2$getRefinedLibSizes(
  clusterType = NULL,
  groups = NULL,
  type = "counts",
  n.cores = self$n.cores
)

Arguments

Returns

recalculated library sizes

Method plotGeneHeatmap()

Plot heatmap for a given set of genes

Usage

Pagoda2$plotGeneHeatmap(
  genes,
  type = "counts",
  clusterType = NULL,
  groups = NULL,
  gradient.range.quantile = 0.95,
  cluster.genes = FALSE,
  inner.clustering = FALSE,
  gradientPalette = NULL,
  v = 0.8,
  s = 1,
  box = TRUE,
  drawGroupNames = FALSE,
  useRaster = TRUE,
  smooth.span = max(1, round(nrow(self$counts)/1024)),
  ...
)

Arguments

Returns

plot of gene heatmap

Method plotEmbedding()

Show embedding

Usage

Pagoda2$plotEmbedding(
  type = NULL,
  embeddingType = NULL,
  clusterType = NULL,
  groups = NULL,
  colors = NULL,
  gene = NULL,
  plot.theme = ggplot2::theme_bw(),
  ...
)

Arguments

Returns

plot of the embedding

Method getOdGenes()

Get overdispersed genes

Usage

Pagoda2$getOdGenes(
  n.odgenes = NULL,
  alpha = 0.05,
  use.unadjusted.pvals = FALSE
)

Arguments

Returns

vector of overdispersed genes

Method getNormalizedExpressionMatrix()

Return variance-normalized matrix for specified genes or a number of OD genes

Usage

Pagoda2$getNormalizedExpressionMatrix(genes = NULL, n.odgenes = NULL)

Arguments

Returns

cell by gene matrix

Method calculatePcaReduction()

Calculate PCA reduction of the data

Usage

Pagoda2$calculatePcaReduction(
  nPcs = 20,
  type = "counts",
  name = "PCA",
  use.odgenes = TRUE,
  n.odgenes = NULL,
  odgenes = NULL,
  center = TRUE,
  cells = NULL,
  fastpath = TRUE,
  maxit = 100,
  verbose = TRUE,
  var.scale = (type == "counts"),
  ...
)

Arguments

Returns

Invisible PCA result (the reduction itself is saved in self$reductions[[name]])"

Method expandOdGenes()

Reset overdispersed genes 'odgenes' to be a superset of the standard odgene selection (guided by n.odgenes or alpha), and a set of recursively determined odgenes based on a given group (or a cluster info)

Usage

Pagoda2$expandOdGenes(
  type = "counts",
  clusterType = NULL,
  groups = NULL,
  min.group.size = 30,
  od.alpha = 0.1,
  use.odgenes = FALSE,
  n.odgenes = NULL,
  odgenes = NULL,
  n.odgene.multiplier = 1,
  gam.k = 10,
  verbose = FALSE,
  n.cores = self$n.cores,
  min.odgenes = 10,
  max.odgenes = Inf,
  recursive = TRUE
)

Arguments

Returns

List of overdispersed genes

Method localPcaKnn()

local PCA implementation

Usage

Pagoda2$localPcaKnn(
  nPcs = 5,
  type = "counts",
  clusterType = NULL,
  groups = NULL,
  k = 30,
  b = 1,
  a = 1,
  min.group.size = 30,
  name = "localPCA",
  od.alpha = 0.1,
  n.odgenes = NULL,
  gam.k = 10,
  verbose = FALSE,
  n.cores = self$n.cores,
  min.odgenes = 5,
  take.top.odgenes = FALSE,
  recursive = TRUE,
  euclidean = FALSE,
  perplexity = k,
  return.pca = FALSE,
  skip.pca = FALSE
)

Arguments

Returns

localPcaKnn return here

Method testPathwayOverdispersion()

Test pathway overdispersion Note: this is a compressed version of the PAGODA1 approach in SCDE <https://hms-dbmi.github.io/scde/>

Usage

Pagoda2$testPathwayOverdispersion(
  setenv,
  type = "counts",
  max.pathway.size = 1000,
  min.pathway.size = 10,
  n.randomizations = 5,
  verbose = FALSE,
  n.cores = self$n.cores,
  score.alpha = 0.05,
  plot = FALSE,
  cells = NULL,
  adjusted.pvalues = TRUE,
  z.score = qnorm(0.05/2, lower.tail = FALSE),
  use.oe.scale = FALSE,
  return.table = FALSE,
  name = "pathwayPCA",
  correlation.distance.threshold = 0.2,
  loading.distance.threshold = 0.01,
  top.aspects = Inf,
  recalculate.pca = FALSE,
  save.pca = TRUE
)

Arguments

Returns

pathway output

Method getEmbedding()

Return embedding

Usage

Pagoda2$getEmbedding(
  type = "counts",
  embeddingType = "largeVis",
  name = NULL,
  dims = 2,
  M = 1,
  gamma = 1/M,
  perplexity = 50,
  verbose = TRUE,
  sgd_batches = NULL,
  diffusion.steps = 0,
  diffusion.power = 0.5,
  distance = "pearson",
  n.cores = self$n.cores,
  n.sgd.cores = n.cores,
  ...
)

Arguments

Returns

embedding stored in self$embedding

Method clone()

The objects of this class are cloneable with this method.

Usage

Pagoda2$clone(deep = FALSE)

Arguments

Author(s)

Simon Steiger

Examples

## ------------------------------------------------
## Method `Pagoda2$new`
## ------------------------------------------------

 
## Load pre-generated a dataset of 50 bone marrow cells as matrix
cm <- readRDS(system.file("extdata", "sample_BM1_50.rds", package="pagoda2"))
## Perform QC, i.e. filter any cells that
counts <- gene.vs.molecule.cell.filter(cm, min.cell.size=500)
rownames(counts) <- make.unique(rownames(counts))
## Generate Pagoda2 object 
p2_object <- Pagoda2$new(counts, log.scale=TRUE, min.cells.per.gene=10, n.cores=1) 



## ------------------------------------------------
## Method `Pagoda2$adjustVariance`
## ------------------------------------------------


## Load pre-generated a dataset of 50 bone marrow cells as matrix
cm <- readRDS(system.file("extdata", "sample_BM1_50.rds", package="pagoda2"))
## Perform QC, i.e. filter any cells that
counts <- gene.vs.molecule.cell.filter(cm, min.cell.size=500)
rownames(counts) <- make.unique(rownames(counts))
## Generate Pagoda2 object 
p2_object <- Pagoda2$new(counts,log.scale=TRUE, min.cells.per.gene=10, n.cores=1) 
## Normalize gene expression variance
p2_object$adjustVariance(plot=TRUE, gam.k=10)



## ------------------------------------------------
## Method `Pagoda2$makeKnnGraph`
## ------------------------------------------------


## Load pre-generated a dataset of 50 bone marrow cells as matrix
cm <- readRDS(system.file("extdata", "sample_BM1_50.rds", package="pagoda2"))
## Perform QC, i.e. filter any cells that
counts <- gene.vs.molecule.cell.filter(cm, min.cell.size=300)
rownames(counts) <- make.unique(rownames(counts))
## Generate Pagoda2 object   
p2_object <- Pagoda2$new(counts,log.scale=TRUE, min.cells.per.gene=10, n.cores=1) 
## Normalize gene expression variance
p2_object$adjustVariance(plot=TRUE, gam.k=10)
## Generate a kNN graph of cells that will allow us to identify clusters of cells
p2_object$makeKnnGraph(k=20, center=FALSE, distance='L2')

Quick utility to check if given character vector is colors Thanks to Stackoverflow: http://stackoverflow.com/questions/13289009/check-if-character-string-is-a-valid-color-representation

Description

Quick utility to check if given character vector is colors Thanks to Stackoverflow: http://stackoverflow.com/questions/13289009/check-if-character-string-is-a-valid-color-representation

Usage

areColors(x)

Arguments

Value

boolean whether character vector is colors

armaCor - matrix column correlations. Allows faster matrix correlations with armadillo. Similar to cor() call, will calculate correlation between matrix columns

Description

armaCor - matrix column correlations. Allows faster matrix correlations with armadillo. Similar to cor() call, will calculate correlation between matrix columns

Usage

armaCor(mat)

Arguments

Value

matrix with columns as correlations

Perform basic 'pagoda2' processing, i.e. adjust variance, calculate pca reduction, make knn graph, identify clusters with multilevel, and generate largeVis and tSNE embeddings.

Description

Perform basic 'pagoda2' processing, i.e. adjust variance, calculate pca reduction, make knn graph, identify clusters with multilevel, and generate largeVis and tSNE embeddings.

Usage

basicP2proc(
  cd,
  n.cores = 1,
  n.odgenes = 3000,
  nPcs = 100,
  k = 30,
  perplexity = 50,
  log.scale = TRUE,
  trim = 10,
  keep.genes = NULL,
  min.cells.per.gene = 0,
  min.transcripts.per.cell = 100,
  get.largevis = TRUE,
  get.tsne = TRUE,
  make.geneknn = TRUE
)

Arguments

Value

a new 'Pagoda2' object

Generate a 'pagoda2' web application from a 'Pagoda2' object

Description

Generate a 'pagoda2' web application from a 'Pagoda2' object

Usage

basicP2web(p2, app.title = "Pagoda2", extraWebMetadata = NULL, n.cores = 4)

Arguments

Value

a 'pagoda2' web object

Rescale the weights in an edge matrix to match a given perplexity. From 'largeVis', <https://github.com/elbamos/largeVis>

Description

Rescale the weights in an edge matrix to match a given perplexity. From 'largeVis', <https://github.com/elbamos/largeVis>

Usage

buildWijMatrix(x, threads = NULL, perplexity = 50)

Arguments

Value

A list with the following components:

'dist'

An [N,K] matrix of the distances to the nearest neighbors.

'id'

An [N,K] matrix of the node indexes of the neartest neighbors. Note that this matrix is 1-indexed, unlike most other matrices in this package.

'k'

The number of nearest neighbors.

Returns a list vector with the number of cells that are present in more than one selections in the provided p2 selection object

Description

Returns a list vector with the number of cells that are present in more than one selections in the provided p2 selection object

Usage

calcMulticlassified(sel)

Arguments

Value

list vector with the number of cells that are present in more than one selections in the provided p2 selection object

Get the number of cells in each selection group

Description

Get the number of cells in each selection group

Usage

cellsPerSelectionGroup(selection)

Arguments

Value

a named vector of cell numbers in each groups

Translate cell cluster dendrogram to an array, one row per node with 1/0 cluster membership

Description

Translate cell cluster dendrogram to an array, one row per node with 1/0 cluster membership

Usage

cldend2array(d, cells = NULL)

Arguments

Value

array with one row per node with 1/0 cluster membership

Collapse aspect patterns into clusters

Description

Collapse aspect patterns into clusters

Usage

collapse.aspect.clusters(d, dw, ct, scale = TRUE, pick.top = FALSE)

Arguments

Value

list of clusters from matrix of normalized aspect patterns and clusters from the corresponding weight matrix

Compare two different clusterings provided as factors by plotting a normalised heatmap

Description

Compare two different clusterings provided as factors by plotting a normalised heatmap

Usage

compareClusterings(cl1, cl2, filename = NA)

Arguments

Value

invisible summary table that gets plotted

Perform differential expression on a p2 object given a set of web selections and two groups to compare

Description

Perform differential expression on a p2 object given a set of web selections and two groups to compare

Usage

diffExprOnP2FromWebSelection(p2, sel, group1name, group2name)

Arguments

Perform differential expression on a p2 object given a set of web selections and one group to compare against everything else

Description

Perform differential expression on a p2 object given a set of web selections and one group to compare against everything else

Usage

diffExprOnP2FromWebSelectionOneGroup(p2, sel, groupname)

Arguments

Perform extended 'Pagoda2' processing. Generate organism specific GO environment and calculate pathway overdispersion.

Description

Perform extended 'Pagoda2' processing. Generate organism specific GO environment and calculate pathway overdispersion.

Usage

extendedP2proc(p2, organism = "hs")

Arguments

Value

list of a 'Pagoda2' object and go.env

Returns a factor of cell membership from a p2 selection object the factor only includes cells present in the selection. If the selection contains multiclassified cells an error is raised

Description

Returns a factor of cell membership from a p2 selection object the factor only includes cells present in the selection. If the selection contains multiclassified cells an error is raised

Usage

factorFromP2Selection(sel, use.internal.name = FALSE, flatten = FALSE)

Arguments

Value

factor of cell membership from a p2 selection object. The factor only includes cells present in the selection.

Converts a list of factors into 'pagoda2' metadata optionally filtering down to the cells present in the provided 'pagoda2' app.

Description

Converts a list of factors into 'pagoda2' metadata optionally filtering down to the cells present in the provided 'pagoda2' app.

Usage

factorListToMetadata(factor.list, p2 = NULL)

Arguments

Value

'pagoda2' web metadata object

Converts a names factor to a p2 selection object if colors are provided it assigns those, otherwise uses a rainbow palette

Description

Converts a names factor to a p2 selection object if colors are provided it assigns those, otherwise uses a rainbow palette

Usage

factorToP2selection(cl, col = NULL)

Arguments

Value

a p2 selection object (list)

Filter cells based on gene/molecule dependency

Description

Filter cells based on gene/molecule dependency

Usage

gene.vs.molecule.cell.filter(
  countMatrix,
  min.cell.size = 500,
  max.cell.size = 50000,
  p.level = min(0.001, 1/ncol(countMatrix)),
  alpha = 0.1,
  plot = TRUE,
  do.par = TRUE
)

Arguments

Value

a filtered matrix

Given a cell clustering (partitioning) and a set of user provided selections generate a cleaned up annotation of cluster groups that can be used for classification

Description

Given a cell clustering (partitioning) and a set of user provided selections generate a cleaned up annotation of cluster groups that can be used for classification

Usage

generateClassificationAnnotation(clustering, selections)

Arguments

Value

a named factor that can be used for classification

Get a control geneset for cell scoring using the method described in Puram, Bernstein (Cell, 2018)

Description

Get a control geneset for cell scoring using the method described in Puram, Bernstein (Cell, 2018)

Usage

get.control.geneset(data, signature, n.bins = 25, n.genes.per.bin = 100)

Arguments

Value

a character vector that can be used as a background signature

Generate differential expression genesets for the web app given a cell grouping by calculating DE sets between each cell set and everything else

Description

Generate differential expression genesets for the web app given a cell grouping by calculating DE sets between each cell set and everything else

Usage

get.de.geneset(pagObj, groups, prefix = "de_")

Arguments

Value

a 'pagoda2' web object

Returns all the cells that are in the designated selections. Given a pagoda2 selections object and the names of some selections in it returns the names of the cells that are in these selections removed any duplicates

Description

Returns all the cells that are in the designated selections. Given a pagoda2 selections object and the names of some selections in it returns the names of the cells that are in these selections removed any duplicates

Usage

getCellsInSelections(p2selections, selectionNames)

Arguments

Value

a character vector of cell names

Assign names to the clusters, given a clustering vector and a set of selections. This function will use a set of pagoda2 cell seletcion to identify the clusters in a a named factor. It is meant to be used to import user defined annotations that are defined as selections into a more formal categorization of cells that are defined by cluster. To help with this the function allows a percent of cells to have been classified in the selections into multiple groups, something which may be the result of the users making wrong selections. The percent of cells allows to be multiselected in any given group is defined by multiClassCutoff. Furthermore the method will assign each cluster to a selection only if the most popular cluster to the next most popular exceed the ambiguous.ratio in terms of cell numbers. If a cluster does not satisfy this condtiion it is not assigned.

Description

Assign names to the clusters, given a clustering vector and a set of selections. This function will use a set of pagoda2 cell seletcion to identify the clusters in a a named factor. It is meant to be used to import user defined annotations that are defined as selections into a more formal categorization of cells that are defined by cluster. To help with this the function allows a percent of cells to have been classified in the selections into multiple groups, something which may be the result of the users making wrong selections. The percent of cells allows to be multiselected in any given group is defined by multiClassCutoff. Furthermore the method will assign each cluster to a selection only if the most popular cluster to the next most popular exceed the ambiguous.ratio in terms of cell numbers. If a cluster does not satisfy this condtiion it is not assigned.

Usage

getClusterLabelsFromSelection(
  clustering,
  selections,
  multiClassCutoff = 0.3,
  ambiguous.ratio = 0.5
)

Arguments

Value

a data.frame with two columns, one for cluster and one for selections, each cluster appears only once

Retrieves the colors of each selection from a p2 selection object as a names vector of strings

Description

Retrieves the colors of each selection from a p2 selection object as a names vector of strings

Usage

getColorsFromP2Selection(sel)

Arguments

Value

a named vector of hex colours

Get a mapping form internal to external names for the specified selection object

Description

Get a mapping form internal to external names for the specified selection object

Usage

getIntExtNamesP2Selection(x)

Arguments

Value

list of names from the specified selection object

Converts the output of hierarchical differential expression aspects into genesets that can be loaded into a 'pagoda2' web app to retrive the genes that make the geneset interactively

Description

Converts the output of hierarchical differential expression aspects into genesets that can be loaded into a 'pagoda2' web app to retrive the genes that make the geneset interactively

Usage

hierDiffToGenesets(output)

Arguments

Value

a geneset that can be loaded into p2 web genesets

Generate a Rook Server app from a 'Pagoda2' object. This generates a 'pagoda2' web object from a 'Pagoda2' object by automating steps that most users will want to run. This function is a wrapper about the 'pagoda2' web constructor. (Advanced users may wish to use that constructor directly.)

Description

Generate a Rook Server app from a 'Pagoda2' object. This generates a 'pagoda2' web object from a 'Pagoda2' object by automating steps that most users will want to run. This function is a wrapper about the 'pagoda2' web constructor. (Advanced users may wish to use that constructor directly.)

Usage

make.p2.app(
  r,
  dendrogramCellGroups,
  additionalMetadata = list(),
  geneSets,
  show.depth = TRUE,
  show.batch = TRUE,
  show.clusters = TRUE,
  appname = "Pagoda2 Application",
  innerOrder = NULL,
  orderDend = FALSE,
  appmetadata = NULL
)

Arguments

Value

a 'pagoda2' web object that presents a Rook compatible interface

Scale the designated values between the range of 0 and 1

Description

Scale the designated values between the range of 0 and 1

Usage

minMaxScale(x)

Arguments

Value

the scaled values

Examples

example_matrix =  matrix(rep(c(1:5), 3), 5)
minMaxScale(example_matrix)

Get a vector of the names of an object named by the names themselves. This is useful with lapply when passing names of objects as it ensures that the output list is also named.

Description

Get a vector of the names of an object named by the names themselves. This is useful with lapply when passing names of objects as it ensures that the output list is also named.

Usage

namedNames(g)

Arguments

Value

vector with names of object

Generate a GO environment for human for overdispersion analysis for the the back end

Description

Generate a GO environment for human for overdispersion analysis for the the back end

Usage

p2.generate.dr.go(r)

Arguments

Value

a GO environment object

Generates zebrafish (Danio rerio) GO annotation for the web object

Description

Generates zebrafish (Danio rerio) GO annotation for the web object

Usage

p2.generate.dr.go.web(gene.names, n.cores = 1)

Arguments

Generate a GO environment for the organism specified

Description

Generate a GO environment for the organism specified

Usage

p2.generate.go(
  r,
  organism = NULL,
  go2all.egs = NULL,
  eg.alias2eg = NULL,
  min.env.length = 5
)

Arguments

Generates GO annotation for the web object for any species

Description

Generates GO annotation for the web object for any species

Usage

p2.generate.go.web(
  gene.names,
  egALIAS2EG = NULL,
  egGO2ALLEGS = NULL,
  n.cores = 1
)

Arguments

Generates GO annotation for the web object from the GO environment used for enrichment analysis

Description

Generates GO annotation for the web object from the GO environment used for enrichment analysis

Usage

p2.generate.go.web.fromGOEnv(go.env)

Arguments

Generate a GO environment for human for overdispersion analysis for the the back end

Description

Generate a GO environment for human for overdispersion analysis for the the back end

Usage

p2.generate.human.go(r)

Arguments

Value

a GO environment object

Generates human GO annotation for the web object

Description

Generates human GO annotation for the web object

Usage

p2.generate.human.go.web(gene.names, n.cores = 1)

Arguments

Generate a GO environment for mouse for overdispersion analysis for the the back end

Description

Generate a GO environment for mouse for overdispersion analysis for the the back end

Usage

p2.generate.mouse.go(r)

Arguments

Value

a GO environment object

Generates mouse (Mus musculus) GO annotation for the web object

Description

Generates mouse (Mus musculus) GO annotation for the web object

Usage

p2.generate.mouse.go.web(gene.names, n.cores = 1)

Arguments

Create 'PAGODA1' web application from a 'Pagoda2' object 'PAGODA1' found here, with 'SCDE': <https://www.bioconductor.org/packages/release/bioc/html/scde.html>

Description

Create 'PAGODA1' web application from a 'Pagoda2' object 'PAGODA1' found here, with 'SCDE': <https://www.bioconductor.org/packages/release/bioc/html/scde.html>

Usage

p2.make.pagoda1.app(
  p2,
  col.cols = NULL,
  row.clustering = NULL,
  title = "pathway clustering",
  zlim = NULL,
  embedding = NULL,
  inner.clustering = TRUE,
  groups = NULL,
  clusterType = NULL,
  embeddingType = NULL,
  veloinfo = NULL,
  type = "PCA",
  min.group.size = 1,
  batch.colors = NULL,
  n.cores = 10
)

Arguments

Value

'PAGODA1' web application

Generate a list metadata structure that can be passed to a 'pagoda2' web object constructor as additional metadata given a named factor

Description

Generate a list metadata structure that can be passed to a 'pagoda2' web object constructor as additional metadata given a named factor

Usage

p2.metadata.from.factor(
  metadata,
  displayname = NULL,
  s = 1,
  v = 1,
  start = 0,
  end = NULL,
  pal = NULL
)

Arguments

Value

list of data, levels, palette to be passed to 'pagoda2' web object constructor

Generate a 'pagoda2' web object from a 'Pagoda2' object using hierarchical differential expression

Description

Generate a 'pagoda2' web object from a 'Pagoda2' object using hierarchical differential expression

Usage

p2.toweb.hdea(p2, title = "")

Arguments

Value

a 'pagoda2' web object

p2ViewPagodaApp R6 class

Description

Modified 'PAGODA1' app (from 'SCDE') for browsing 'pagoda2' results. Refer to 'ViewPagodaAppOld' and 'make.pagoda.app()' in 'SCDE'

Public fields

Methods

Public methods

• p2ViewPagodaApp$new()

• p2ViewPagodaApp$getgenecldata()

• p2ViewPagodaApp$call()

• p2ViewPagodaApp$clone()

Method new()

Initialize p2ViewPagodaApp class

Usage

p2ViewPagodaApp$new(
  results,
  pathways,
  genes,
  goenv,
  batch = NULL,
  name = "pathway overdispersion",
  trim = 1.1/nrow(p2$counts),
  embedding = NULL,
  type,
  veloinfo = NULL
)

Arguments

Returns

new 'p2ViewPagodaApp' object

Method getgenecldata()

Helper function to get the heatmap data for a given set of genes

Usage

p2ViewPagodaApp$getgenecldata(genes = NULL, gcl = NULL, ltrim = 0)

Arguments

Returns

heatmap data for a given set of genes

Method call()

Call Rook application. Using client-side ExtJS framework and Inchlib HTML5 canvas libraries to create the graphical user interface for PAGODA

Usage

p2ViewPagodaApp$call(env)

Arguments

Returns

modified 'PAGODA1' app

Method clone()

The objects of this class are cloneable with this method.

Usage

p2ViewPagodaApp$clone(deep = FALSE)

Arguments

Collapse aspects driven by the same combinations of genes. (Aspects are some pattern across cells e.g. sequencing depth, or PC corresponding to an undesired process such as ribosomal pathway variation.) Examines PC loading vectors underlying the identified aspects and clusters of aspects based on a product of loading and score correlation (raised to corr.power). Clusters of aspects driven by the same genes are determined based on the parameter "distance.threshold".

Description

Collapse aspects driven by the same combinations of genes. (Aspects are some pattern across cells e.g. sequencing depth, or PC corresponding to an undesired process such as ribosomal pathway variation.) Examines PC loading vectors underlying the identified aspects and clusters of aspects based on a product of loading and score correlation (raised to corr.power). Clusters of aspects driven by the same genes are determined based on the parameter "distance.threshold".

Usage

pagoda.reduce.loading.redundancy(
  tam,
  pwpca,
  clpca = NULL,
  plot = FALSE,
  cluster.method = "complete",
  distance.threshold = 0.01,
  corr.power = 4,
  abs = TRUE,
  n.cores = 1,
  ...
)

Arguments

Value

a list structure analogous to that returned by pagoda.top.aspects(), but with addition of a $cnam element containing a list of aspects summarized by each row of the new (reduced) $xv and $xvw

Collapse aspects driven by similar patterns (i.e. separate the same sets of cells) Examines PC loading vectors underlying the identified aspects and clusters aspects based on score correlation. Clusters of aspects driven by the same patterns are determined based on the distance.threshold.

Description

Collapse aspects driven by similar patterns (i.e. separate the same sets of cells) Examines PC loading vectors underlying the identified aspects and clusters aspects based on score correlation. Clusters of aspects driven by the same patterns are determined based on the distance.threshold.

Usage

pagoda.reduce.redundancy(
  tamr,
  distance.threshold = 0.2,
  cluster.method = "complete",
  distance = NULL,
  weighted.correlation = TRUE,
  plot = FALSE,
  top = Inf,
  trim = 0,
  abs = FALSE,
  ...
)

Arguments

Value

List structure analogous to that returned by pagoda.top.aspects(), but with addition of a $cnam element containing a list of aspects summarized by each row of the new (reduced) $xv and $xvw

pagoda2WebApp class to create 'pagoda2' web applications via a Rook server

Description

pagoda2WebApp class to create 'pagoda2' web applications via a Rook server

Fields

originalP2object

Input 'Pagoda2' object

name

string Display name for the application

mat

Embedding

cellmetadata

Metadata associated with 'Pagoda2' object

mainDendrogram

Dendrogram from hclust() of all cells in the 'Pagoda2' object

geneSets

Gene sets in the 'Pagoda2' object

rookRoot

Rook server root directory

appmetadata

pagoda2 web application metadata

pagoda2WebApp_arrayToJSON

Description

Serialise an R array to a JSON object

Arguments

Value

Serialised version of the array in JSON, which includes dimension information as separate fields

pagoda2WebApp_availableAspectsJSON

Description

Parse pathways from originalP2object$misc$pathwayOD$xv into JSON

Value

JSON with parsed cell order from mainDendrogram$cellorder

pagoda2WebApp_call

Description

Handle httpd server calls

Arguments

pagoda2WebApp_cellOrderJSON

Description

Parse mainDendrogram$cellorder into JSON

Value

JSON with parsed cell order from mainDendrogram$cellorder

pagoda2WebApp_cellmetadataJSON

Description

Parse cellmetadata into JSON

Value

JSON with parsed cellmetadata

pagoda2WebApp_geneInformationJSON

Description

Parse originalP2object$misc$varinfo[,c("m","qv")] into JSON

Value

JSON with parsed information from genename, dispersion, mean gene expression

Generate a dendrogram of groups

Description

Generate a dendrogram of groups

Arguments

Value

List of hcGroups, cellorder, and cluster.sizes

pagoda2WebApp_generateEmbeddingStructure

Description

Generate information about the embeddings we are exporting

Value

List with embeddings

pagoda2WebApp_generateGeneKnnJSON

Description

Generate a JSON list representation of the gene kNN network

Arguments

Value

JSON with gene kNN network

pagoda2WebApp_getCompressedEmbedding

Description

Compress the embedding

Arguments

Value

compressed embedding as JSON

pagoda2WebApp_packCompressFloat64Array

Description

Compress float64 array

Arguments

Value

compressed array

pagoda2WebApp_packCompressInt32Array

Description

Compress int32 array

Arguments

Value

compressed array

pagoda2WebApp_readStaticFile

Description

Read a static file from the filesystem, and put in the response

Arguments

Value

Content to display or error page

pagoda2WebApp_reducedDendrogramJSON

Description

Parse dendrogram into JSON

Value

JSON with parsed dendrogram

pagoda2WebApp_serializeToStaticFast

Description

Convert serialized file to static file

Arguments

Value

static file written by WriteListToBinary(expL=exportList, outfile=binary.filename, verbose=verbose)

pagoda2WebApp_serverLog

Description

Logging function for console

Arguments

Value

printed message

pagoda2WebApp_sparseMatList

Description

Create simple List from sparse Matrix with Dimnames as JSON

Arguments

Value

List with slots i, p, x

Parallel, optionally verbose lapply. See ?parallel::mclapply for more info.

Description

Parallel, optionally verbose lapply. See ?parallel::mclapply for more info.

Usage

papply(..., n.cores = parallel::detectCores(), mc.preschedule = FALSE)

Arguments

Value

list, as returned by lapply

Calculate correlation distance between PC magnitudes given a number of target dimensions

Description

Calculate correlation distance between PC magnitudes given a number of target dimensions

Usage

pathway.pc.correlation.distance(pcc, xv, n.cores = 1, target.ndf = NULL)

Arguments

Value

correlation distance matrix, akin to stats dist

Plot multiclassified cells per selection as a percent barplot

Description

Plot multiclassified cells per selection as a percent barplot

Usage

plotMulticlassified(sel)

Arguments

Value

ggplot2 object

Plot the embedding of a 'Pagoda2' object with the given values

Description

Plot the embedding of a 'Pagoda2' object with the given values

Usage

plotOneWithValues(
  p2obj,
  values,
  title = "",
  type = "PCA",
  embeddingType = "tSNE"
)

Arguments

Value

NULL, simply updates p2obj$plotEmbedding()

Get a dataframe and plot summarising overlaps between selection of a pagoda2 selection object ignore self overlaps

Description

Get a dataframe and plot summarising overlaps between selection of a pagoda2 selection object ignore self overlaps

Usage

plotSelectionOverlaps(sel)

Arguments

Value

a list that contains a ggplot2 object and a datatable with the overlaps data

Project a distance matrix into a lower-dimensional space. (from elbamos/largeVis)

Description

Takes as input a sparse matrix of the edge weights connecting each node to its nearest neighbors, and outputs a matrix of coordinates embedding the inputs in a lower-dimensional space.

Usage

projectKNNs(
  wij,
  dim = 2,
  sgd_batches = NULL,
  M = 5,
  gamma = 7,
  alpha = 1,
  rho = 1,
  coords = NULL,
  useDegree = FALSE,
  momentum = NULL,
  seed = NULL,
  threads = NULL,
  verbose = getOption("verbose", TRUE)
)

Arguments

Details

The algorithm attempts to estimate a dim-dimensional embedding using stochastic gradient descent and negative sampling.

The objective function is:

O = \sum_{(i,j)\in E} w_{ij} (\log f(||p(e_{ij} = 1||) + \sum_{k=1}^{M} E_{jk~P_{n}(j)} \gamma \log(1 - f(||p(e_{ij_k} - 1||)))

where f() is a probabilistic function relating the distance between two points in the low-dimensional projection space, and the probability that they are nearest neighbors.

The default probabilistic function is 1 / (1 + \alpha \dot ||x||^2). If \alpha is set to zero, an alternative probabilistic function, 1 / (1 + \exp(x^2)) will be used instead.

Note that the input matrix should be symmetric. If any columns in the matrix are empty, the function will fail.

Value

A dense [N,D] matrix of the coordinates projecting the w_ij matrix into the lower-dimensional space.

Note

If specified, seed is passed to the C++ and used to initialize the random number generator. This will not, however, be sufficient to ensure reproducible results, because the initial coordinate matrix is generated using the R random number generator. To ensure reproducibility, call set.seed before calling this function, or pass it a pre-allocated coordinate matrix.

The original paper called for weights in negative sampling to be calculated according to the degree of each vertex, the number of edges connecting to the vertex. The reference implementation, however, uses the sum of the weights of the edges to each vertex. In experiments, the difference was imperceptible with small (MNIST-size) datasets, but the results seems aesthetically preferrable using degree. The default is to use the edge weights, consistent with the reference implementation.

Quick loading of 10X CellRanger count matrices

Description

Quick loading of 10X CellRanger count matrices

Usage

read.10x.matrices(matrixPaths, version = "V3", n.cores = 1, verbose = TRUE)

Arguments

Value

a sparse matrix representation of the data (or a list of sparse matrices if a list of paths was passed)

This function reads a matrix generated by the 10x processing pipeline from the specified directory and returns it. It aborts if one of the required files in the specified directory do not exist.

Description

This function reads a matrix generated by the 10x processing pipeline from the specified directory and returns it. It aborts if one of the required files in the specified directory do not exist.

Usage

read10xMatrix(path, version = "V3", transcript.id = "SYMBOL", verbose = TRUE)

Arguments

Value

parsed 10x outputs into a matrix

Read a pagoda2 cell selection file and return it as a factor while removing any mutliclassified cells

Description

Read a pagoda2 cell selection file and return it as a factor while removing any mutliclassified cells

Usage

readPagoda2SelectionAsFactor(filepath, use.internal.name = FALSE)

Arguments

Value

a name factor with the membership of all the cells that are not multiclassified

Reads a 'pagoda2' web app exported cell selection file exported as a list of list objects that contain the name of the selection, the color (as a hex string) and the identifiers of the individual cells

Description

Reads a 'pagoda2' web app exported cell selection file exported as a list of list objects that contain the name of the selection, the color (as a hex string) and the identifiers of the individual cells

Usage

readPagoda2SelectionFile(filepath)

Arguments

Remove cells that are present in more than one selection from all the selections they are in

Description

Remove cells that are present in more than one selection from all the selections they are in

Usage

removeSelectionOverlaps(selections)

Arguments

Value

a new list with the duplicated cells removed

Score cells by getting mean expression of genes in signatures

Description

Score cells by getting mean expression of genes in signatures

Usage

score.cells.nb0(data, signature)

Arguments

Value

cell scores

Score cells after standardising the expression of each gene removing outliers

Description

Score cells after standardising the expression of each gene removing outliers

Usage

score.cells.nb1(data, signature, quantile.cutoff = 0.01)

Arguments

Value

The filtered subset of gene signatures

Puram, Bernstein (Cell, 2018) Score cells as described in Puram, Bernstein (Cell, 2018)

Description

Puram, Bernstein (Cell, 2018) Score cells as described in Puram, Bernstein (Cell, 2018)

Usage

score.cells.puram(data, signature, correct = TRUE, show.plot = FALSE, ...)

Arguments

Value

a score for each cell

Calculate the default number of batches for a given number of vertices and edges. The formula used is the one used by the 'largeVis' reference implementation. This is substantially less than the recommendation E * 10000 in the original paper.

Description

Calculate the default number of batches for a given number of vertices and edges. The formula used is the one used by the 'largeVis' reference implementation. This is substantially less than the recommendation E * 10000 in the original paper.

Usage

sgdBatches(N, E = 150 * N/2)

Arguments

Value

The recommended number of sgd batches.

Examples

# Observe that increasing K has no effect on processing time

N <- 70000 # MNIST
K <- 10:250
plot(K, sgdBatches(rep(N, length(K)), N * K / 2))

# Observe that processing time scales linarly with N
N <- c(seq(from = 1, to = 10000, by = 100), seq(from = 10000, to = 10000000, by = 1000))
plot(N, sgdBatches(N))

Directly open the 'pagoda2' web application and view the 'p2web' application object from our R session

Description

Directly open the 'pagoda2' web application and view the 'p2web' application object from our R session

Usage

show.app(app, name, port, ip, browse = TRUE, server = NULL)

Arguments

Value

application within browser

Set names equal to values, a stats::setNames wrapper function

Description

Set names equal to values, a stats::setNames wrapper function

Usage

sn(x)

Arguments

Value

An object with names assigned equal to values

Subset a gene signature to the genes in the given matrix with optional warning if genes are missing

Description

Subset a gene signature to the genes in the given matrix with optional warning if genes are missing

Usage

subsetSignatureToData(data, signature, raise.warning = TRUE)

Arguments

Value

The filtered subset of gene signatures

View pathway or gene-weighted PCA 'Pagoda2' version of the function pagoda.show.pathways() Takes in a list of pathways (or a list of genes), runs weighted PCA, optionally showing the result.

Description

View pathway or gene-weighted PCA 'Pagoda2' version of the function pagoda.show.pathways() Takes in a list of pathways (or a list of genes), runs weighted PCA, optionally showing the result.

Usage

tp2c.view.pathways(
  pathways,
  p2,
  goenv = NULL,
  batch = NULL,
  n.genes = 20,
  two.sided = TRUE,
  n.pc = rep(1, length(pathways)),
  colcols = NULL,
  zlim = NULL,
  labRow = NA,
  vhc = NULL,
  cexCol = 1,
  cexRow = 1,
  nstarts = 50,
  row.order = NULL,
  show.Colv = TRUE,
  plot = TRUE,
  trim = 1.1/nrow(p2$counts),
  showPC = TRUE,
  ...
)

Arguments

Value

cell scores along the first principal component of shown genes (returned as invisible)

Validates a pagoda2 selection object

Description

Validates a pagoda2 selection object

Usage

validateSelectionsObject(selections)

Arguments

Value

a logical value indicating if the object is valid

Internal function to visualize aspects of transcriptional heterogeneity as a heatmap.

Description

Internal function to visualize aspects of transcriptional heterogeneity as a heatmap.

Usage

view.aspects(
  mat,
  row.clustering = NA,
  cell.clustering = NA,
  zlim = c(-1, 1) * quantile(mat, p = 0.95),
  row.cols = NULL,
  col.cols = NULL,
  cols = colorRampPalette(c("darkgreen", "white", "darkorange"), space = "Lab")(1024),
  show.row.var.colors = TRUE,
  top = Inf,
  ...
)

Arguments

Value

A heatmap

Generate a 'pagoda2' web object

Description

Generate a 'pagoda2' web object

Usage

webP2proc(
  p2,
  additionalMetadata = NULL,
  title = "Pagoda2",
  make.go.sets = TRUE,
  make.de.sets = TRUE,
  go.env = NULL,
  make.gene.graph = TRUE,
  appmetadata = NULL
)

Arguments

Value

a 'pagoda2' web application

Sets the ncol(mat)*trim top outliers in each row to the next lowest value same for the lowest outliers

Description

Sets the ncol(mat)*trim top outliers in each row to the next lowest value same for the lowest outliers

Usage

winsorize.matrix(mat, trim)

Arguments

Value

Winsorized matrix

Examples

set.seed(0)
mat <- matrix( c(rnorm(5*10,mean=0,sd=1), rnorm(5*10,mean=5,sd=1)), 10, 10)  # random matrix
mat[1,1] <- 1000  # make outlier
range(mat)  # look at range of values
win.mat <- winsorize.matrix(mat, 0.1)
range(win.mat)  # note outliers removed

Writes a list of genes as a gene selection that can be loaded in the web interface

Description

Writes a list of genes as a gene selection that can be loaded in the web interface

Usage

writeGenesAsPagoda2Selection(name, genes, filename)

Arguments

Value

NULL, writes to filepath the list of genes as a gene selection that can be loaded in the web interface

Writes a pagoda2 selection object as a p2 selection file that be be loaded to the web interface

Description

Writes a pagoda2 selection object as a p2 selection file that be be loaded to the web interface

Usage

writePagoda2SelectionFile(sel, filepath)

Arguments

Value

NULL, writes to filepath the pagoda2 selection object as a p2 selection file that be be loaded to the web interface