A plot-friendly output for the Barnes-Hut implementation of t-SNE: test different perplexities and dimensions. See Rtsne::Rtsne for details on the algorithm.
matrix; each row is an observation (eg. cell); each column is a feature (eg. gene / principal component)
numeric vector of perplexities to use. Default: c(15, 30, 50)
numeric vector of dimensions to use. Default: ncol(m)
other arguments passed to Rtsne::Rtsne
a dataframe with columns: id, tSNE1, tSNE2, perplexity, ndim
Every combination of values for 'perplexity' and 'ndim' will be run if multiple are provided for each.