spectral_tsne: Reduce dimensions of a matrix with pca and tSNE
In jacobheng/cellwrangler: Supplemental toolkit for single-cell RNAseq analysis

Description Usage Arguments Value Examples

spectral_tsne() performs dimensionality reduction using the package Rtsne's wrapper for the C++ implementation of Barnes-Hut t-Distributed Stochastic Neighbor Embedding. If no prcomp_object is supplied, principal component analysis will be performed using an irlba algorithm for sparse data.

1 2	spectral_tsne(matrix, log_matrix = TRUE, prcomp_object = NULL, dims = 1:10, perplexity = 30)

`matrix`	a matrix of values to perform dimensionality reduction on; by default, rows are genes and columns are cells
`log_matrix`	if log10 transformation is to be performed on the matrix; defaults to TRUE
`dims`	dimensions from the prinicpal component analysis to use; defaults to 1:10 (i.e. 1st to 10th principal components)
`perplexity`	numeric; perplexity parameter for tSNE; defaults to 30
`prcomp_obj`	a principal component analysis object produced by the prcomp or prcomp_irlba functions; if no object is supplied, sparse_pca will be run on the matrix to return 50 dimensions; defaults to NULL; if a prcomp object is supplied, matrix is not required
`pca_version`	PCA implementation to use. Possible values are "default" for sparse_pca() or "monocle" for the sparse_irlba_prcomp implemented in Monocle 3 alpha.
`center`	a logical value indicating whether the variables should be shifted to be zero centered. Alternately, if the "monocle" pca version is used, a centering vector of length equal the number of columns of x can be supplied.
`scale`	a logical value indicating whether the variables should be scaled to have unit variance before the analysis takes place. If the "default" pca version is used and center = TRUE, scaling will be also default to TRUE. Alternatively, if the "monocle" pca version is used, a vector of length equal the number of columns of x can be supplied.