nearest_neighbour_pipeline: Nearest neighbour pipeline

In wikiselev/clustools: Tools for cluster analysis of single cell high throughput data

Description

Performs and evaluates kmeans clustering with a given combination of gene filtering 2, distance metrics, transformation, number of dimensions used in clustering and number of nearest neighbours.

Usage

nearest_neighbour_pipeline(dataset, sel, distan, clust, n.dim, nn)

Arguments

dataset	Name of the dataset. Either "quake", "sandberg", "bernstein" or "linnarsson"
sel	Selection method used by gene_filter2() function (either "none", "correlation", "variance", "variance_weight", "shannon_weight")
distan	Distance metrics for calculating a distance matrix (either "pearson", "spearman", "euclidean", "manhattan" or "minkowski").
clust	Distance matrix transformation method (either "pca", "spectral", "spectral_reg" or "mds")
n.dim	Number of dimension of the transformed distance matrix which is used in kmeans clustering.
nn	Number of nearest neighbours used

Value

Two small files (depending on the transformation method) containing a set of clustering evaluation indecies (*-inds.txt) together with known and calculated clustering labels of the cells (*-labs.txt). Evaluation indecies are the following:

ari

Adjusted Rand Index

rand

Rand Index

jaccard

Jaccard Index

dunn

Dunn Index

davies_bouldin

Davies Bouldin Index

silhouette

Silhouette Index

Examples

nearest_neighbour_pipeline("quake", "none", "spearman", "spectral", 4, 3)

wikiselev/clustools documentation built on May 4, 2019, 5:25 a.m.