ti_embeddr: Embeddr

View source: R/ti_embeddr.R

ti_embeddrR Documentation

Embeddr

Description

Will generate a trajectory using Embeddr.

This method was wrapped inside a container. The original code of this method is available here.

Usage

ti_embeddr(
  ndim = 2L,
  kernel = "nn",
  metric = "correlation",
  nn_pct = 0L,
  eps = 0L,
  t = 0L,
  symmetrize = "mean",
  measure_type = "unorm",
  thresh = 0.001,
  maxit = 10L,
  stretch = 2L,
  smoother = "smooth.spline"
)

Arguments

ndim

Dimension of the embedded space, default is 2. Domain: U(2, 10). Default: 2. Format: integer.

kernel

The choice of kernel. 'nn' will give nearest neighbours, 'dist' gives minimum distance and'heat' gives a heat kernel. Discussed in detail in 'Laplacian Eigenmaps and Spectral Techniques for Embedding and Clustering',Belkin & Niyogi. Domain: nn, dist, heat. Default: nn. Format: character.

metric

The metric with which to assess 'closeness' for nearest neighbour selection, one of'correlation' (pearson) or 'euclidean'. Default is 'correlation'. Domain: correlation, euclidean, cosine. Default: correlation. Format: character.

nn_pct

The percentage of cells to use as tge number of nearest neighbours if kernel == 'nn'. Domain: U(-2, 1). Default: 0. Format: numeric.

eps

Maximum distance parameter if kernel == 'dist'. Domain: U(-5, 5). Default: 0. Format: numeric.

t

"time" for heat kernel if kernel == "heat". Domain: U(-5, 5). Default: 0. Format: numeric.

symmetrize

How to make the adjacency matrix symmetric. Note that slightlycounterintuitively, node i having node j as a nearest neighbour doesn't guarantee nodej has node i. There are several ways to get round this;* mean If the above case occurs make the link weight 0.5 so the adjacency matrix becomes $0.5(A + A')$* ceil If the above case occurs set the link weight to 1 (ie take the ceiling of the mean case)* floor If the above case occurs set the link weight to 0 (ie take the floor of the mean case). Domain: mean, ceil, floor. Default: mean. Format: character.

measure_type

Type of laplacian eigenmap, which corresponds to the constraint on the eigenvalue problem. Iftype is 'unorm' (default), then the graph measure used is the identity matrix, while if type is 'norm' then the measureused is the degree matrix. Domain: unorm, norm. Default: unorm. Format: character.

thresh

Convergence threshold on shortest distances to the curve. Domain: e^U(-11.51, 11.51). Default: 0.001. Format: numeric.

maxit

Maximum number of iterations. Domain: U(0, 50). Default: 10. Format: integer.

stretch

A factor by which the curve can be extrapolated when points are projected. Default is 2 (times the last segment length). The default is 0 for smoother equal to "periodic_lowess". Domain: U(0, 5). Default: 2. Format: numeric.

smoother

Choice of smoother. The default is "smooth_spline", and other choices are "lowess" and "periodic_lowess". The latter allows one to fit closed curves. Beware, you may want to use iter = 0 with lowess(). Domain: smooth.spline, lowess, periodic.lowess. Default: smooth.spline. Format: character.

Value

A TI method wrapper to be used together with infer_trajectory

References

Campbell, K., Ponting, C.P., Webber, C., 2015. Laplacian eigenmaps and principal curves for high resolution pseudotemporal ordering of single-cell RNA-seq profiles.


dynverse/dynmethods documentation built on Jan. 18, 2024, 4:44 a.m.