run_tSpace: Run the matlab version of the tSpace algorithm

In kbrulois/MATLABfuncs: R interface for the MATLAB implementation of the tSpace and wanderlust algorithms

Description

This is an interface to the matlab implementation of the tSpace algorithm described in Dermadi et al, modified for improved error handling.

Usage

run_tSpace(
  data,
  trajectories = 100,
  k = c(20, 30, 50, 100),
  l = "auto",
  metric = "euclidean",
  graphs = 5,
  landmarks = 20,
  voting_scheme = "exponential",
  label = "",
  matlab_version = "auto",
  matlab_path = "~/Documents/MATLAB"
)

Arguments

data	A matrix-like object where cells correspond to rows and features correspond to columns
trajectories	The number of trajectories to calculate.
k	Number of nearest neighbors (passed to the k argument of wanderlust). If a vector of integers is provided (recommended), the tSpace calculation will be attempted on the smallest value in the vector and then reattempted on successively higher values of k should the algorithm fail. tSpace caclualtions that require an excessively high k value suggest the presence of outlier data points that should be removed from the input data for better performance.
l	Number of nearest neighbors (passed to the l argumet of wanderlust). If set to 'auto' (default), this parameter will be 2/3 of k (rounded to the nearest integer). If k is a vector, l must be a vector the same length and all corresponding values must be less than k.
metric	A distance metric accepted by the dist function in matlab (wanderlust): euclidean, correlation, cityblock, chebyshev, minkowski, hamming, mahalanobis, jaccard, cosine or spearman.
graphs	Number of graphs over which to average trajectories (wanderlust).
landmarks	Number of landmarks (wanderlust).
voting_scheme	Arugment passed to wanderlust.
label	A character string to be incorporated into the tSpace column labels. Default is an empty label.
matlab_version	A character string specifying the active matlab version (e.g. "R2019a"). If set to "auto," the latest version of matlab will be used. Only works for mac users.
matlab_path	The path to the MATLAB folder containing cyt and tSpace scripts.

Value

A list that includes the following elements:

tPCs

A cell by tPC matrix of the PCA reduction of the trajectory data.

traj

A cell by trajectory matrix containing the trajectories.

clusters

A vector containing the cluster definitions used to initiate tSpace.

pPCA

A cell by pPCA matrix of the PCA reduction of the parameters.

Author(s)

Kevin Brulois

kbrulois/MATLABfuncs documentation built on Sept. 14, 2021, 2:02 p.m.