RunSWNE: Wrapper for running SWNE analysis

In yanwu2014/swne: Similarity Weighted Nonnegative Embedding: A method for visualizing high dimensional datasets

Wrapper for running SWNE analysis

Description

Wrapper for running SWNE analysis

Usage

## S3 method for class 'cisTopic'
RunSWNE(
  cisTopicObject,
  proj.method = "sammon",
  cells.use = NULL,
  dist.metric = "cosine",
  n.cores = 8,
  hide.factors = T,
  n_pull = 3,
  alpha.exp = 1.25,
  snn.exp = 1,
  snn.k = 20,
  prune.SNN = 1/15,
  use.paga.pruning = T,
  sample.groups = NULL,
  paga.qval.cutoff = 0.001
)

## S3 method for class 'Seurat'
RunSWNE(
  object,
  proj.method = "sammon",
  reduction.use = "pca",
  cells.use = NULL,
  dims.use = NULL,
  genes.use = NULL,
  dist.metric = "cosine",
  distance.matrix = NULL,
  n.cores = 8,
  k,
  k.range,
  var.genes,
  loss = "mse",
  genes.embed,
  hide.factors = T,
  n_pull = 3,
  ica.fast = T,
  alpha.exp = 1.25,
  snn.exp = 1,
  snn.k = 10,
  use.paga.pruning = T,
  sample.groups = NULL,
  paga.qval.cutoff = 0.001,
  reduction.name = "swne",
  reduction.key = "SWNE_",
  return.format = "embedding",
  ...
)

## S3 method for class 'Pagoda2'
RunSWNE(
  object,
  proj.method = "sammon",
  dist.metric = "cosine",
  n.cores = 8,
  k,
  k.range,
  var.genes,
  loss = "mse",
  genes.embed,
  hide.factors = T,
  n_pull = 3,
  ica.fast = T,
  n.var.genes = 3000,
  alpha.exp = 1.25,
  snn.exp = 1,
  snn.k = 20,
  use.paga.pruning = T,
  sample.groups = NULL,
  paga.qval.cutoff = 0.001
)

## S3 method for class 'dgCMatrix'
RunSWNE(
  data.matrix,
  proj.method = "sammon",
  dist.metric = "cosine",
  n.cores = 3,
  k,
  k.range,
  var.genes,
  loss = "mse",
  genes.embed,
  hide.factors = T,
  n_pull = 3,
  ica.fast = T,
  alpha.exp = 1.25,
  snn.exp = 1,
  snn.k = 20,
  use.paga.pruning = T,
  sample.groups = NULL,
  paga.qval.cutoff = 0.001
)

## S3 method for class 'matrix'
RunSWNE(
  data.matrix,
  proj.method = "sammon",
  dist.metric = "cosine",
  n.cores = 8,
  k,
  k.range,
  var.genes = rownames(data.matrix),
  loss = "mse",
  genes.embed,
  hide.factors = T,
  n_pull = 3,
  alpha.exp = 1.25,
  snn.exp = 1
)

## S3 method for class 'dgTMatrix'
RunSWNE(
  data.matrix,
  proj.method = "sammon",
  dist.metric = "cosine",
  n.cores = 3,
  k,
  k.range,
  var.genes = rownames(data.matrix),
  loss = "mse",
  genes.embed,
  hide.factors = T,
  n_pull = 3,
  alpha.exp = 1.25,
  snn.exp = 1
)

Arguments

proj.method	Method to use to project factors in 2D. Either "sammon" or "umap"
cells.use	Which cells to analyze (default, all cells)
n.cores	Number of cores to use (passed to FindNumFactors)
hide.factors	Hide factors when plotting SWNE embedding
n_pull	Maximum number of factors "pulling" on each sample
alpha.exp	Increasing alpha.exp increases how much the NMF factors "pull" the samples (passed to EmbedSWNE)
snn.exp	Decreasing snn.exp increases the effect of the similarity matrix on the embedding (passed to EmbedSWNE)
snn.k	Changes the number of nearest neighbors used to build SNN (passed to CalcSNN)
prune.SNN	The minimum fraction of shared nearest neighbors (smaller values are set to zero)
use.paga.pruning	Use PAGA graphs to prune
sample.groups	Clusters to use for PAGA (default is to do a de-novo clustering)
paga.qval.cutoff	q-value cutoff for significant shared edges between clusters
object	A Seurat, Pagoda2 or cisTopicObject object with normalized data
reduction.use	Which dimensional reduction (e.g. PCA, ICA) to use for the tSNE. Default is PCA.
dims.use	Which dimensions to use as input features
genes.use	If set, run the SWNE on this subset of genes (instead of running on a set of reduced dimensions). Not set (NULL) by default
distance.matrix	If set, runs tSNE on the given distance matrix instead of data matrix (experimental)
k	Number of NMF factors (passed to RunNMF). If none given, will be derived from FindNumFactors.
k.range	Range of factors for FindNumFactors to iterate over if k is not given
var.genes	vector to specify variable genes. Will infer from Seurat or use full dataset if not given.
loss	loss function to use (passed to RunNMF)
genes.embed	Genes to add to the SWNE embedding
ica.fast	Whether to run SVD before ICA initialization
reduction.name	dimensional reduction name, specifies the position in the object$dr list. swne by default
reduction.key	dimensional reduction key, specifies the string before the number for the dimension names. SWNE_ by default
return.format	format to return ("seurat" object or raw "embedding")
n.var.genes	Number of variable genes to use
data.matrix	a data matrix (genes x cells) which has been pre-normalized
batch	Vector of batch IDs to regress away
dist.use	Similarity function to use for calculating factor positions (passed to EmbedSWNE). Options include pearson (correlation), IC (mutual information), cosine, euclidean.

Value

A list of factor (H.coords) and sample coordinates (sample.coords) in 2D

yanwu2014/swne documentation built on Aug. 5, 2023, 4:42 a.m.