clustering_angular_distance: clustering_angular_distance

In MitoHEAR: Quantification of Mitochondrial DNA Heteroplasmy

clustering_angular_distance

Description

For each pair of samples and for each base, an angular distance matrix is computed based on the four allele frequencies. Then only the angular distances corresponding to the relevant_bases are kept. If relevant bases is NULL, then only the angular distances corresponding to the bases with relative distance variance among samples above min_value are kept . Finally the distance between each pair of samples is defined as the euclidean distance of the angular distances corresponding to the bases that pass the previous filtering step. On this final distance matrix, a hierarchical clustering approach is performed using the function cutreeHybrid of the package dynamicTreeCut.

Usage

clustering_angular_distance(
  heteroplasmy_matrix,
  allele_matrix,
  cluster,
  top_pos,
  deepSplit_param,
  minClusterSize_param,
  threshold = 0.2,
  min_value,
  index,
  relevant_bases = NULL,
  max_frac = 0.7
)

Arguments

heteroplasmy_matrix	Third element returned by get_heteroplasmy.
allele_matrix	Fourth element returned by get_heteroplasmy.
cluster	Vector specifying a partition of the samples.
top_pos	Numeric value. Number of bases sorted with decreasing values of distance variance (see section Details below) among samples. If relevant_bases=NULL, then the bases for performing hierarchical clustering are the ones whose relative variance (variance of the base divided sum of variance among top_pos bases) is above min_value.
deepSplit_param	Integer value between 0 and 4 for the deepSplit parameter of the function cutreeHybrid. See section Details below.
minClusterSize_param	Integer value specifying the minClusterSize parameter of the function cutreeHybrid. See section Details below.
threshold	Numeric value. If a base has heteroplasmy greater or equal to threshold in more than max_frac of cells, then the base is not considered for down stream analysis.
min_value	Numeric value. If relevant_bases=NULL, then the bases for performing hierarchical clustering are the ones whose relative variance (variance of the base divided sum of variance among top_pos bases) is above min_value.
index	Fifth element returned by get_heteroplasmy.
relevant_bases	Character vector of bases to consider as features for performing hierarchical clustering on samples.Default=NULL.
max_frac	Numeric value.If a base has heteroplasmy greater or equal to threshold in more than max_frac of cells, then the base is not considered for down stream analysis.

Value

It returns a list with 4 elements:

classification	Dataframe with two columns and n_row equal to n_row in heteroplasmy_matrix. The first column is the old cluster annotation provided by cluster. The second columns is the new cluster annotation obtained with hierarchical clustering on distance matrix based on heteroplasmy values.
dist_ang_matrix	Distance matrix based on heteroplasmy values as defined in the section Details
top_bases_dist	Vector of bases used for hierarchical clustering. If relevant_bases is not NULL, then top_bases_dist=NULL
common_idx	Vector of indices of samples for which hierarchical clustering is performed. If index is NULL, then common_idx=NULL

Author(s)

Gabriele Lubatti [e-mail]

See Also

https://www.rdocumentation.org/packages/dynamicTreeCut/versions/1.63-1/topics/cutreeHybrid

MitoHEAR documentation built on March 18, 2022, 6:47 p.m.