pagoda.reduce.loading.redundancy {pagoda2}

Collapse aspects driven by the same combinations of genes. (Aspects are some pattern across cells e.g. sequencing depth, or PC corresponding to an undesired process such as ribosomal pathway variation.) Examines PC loading vectors underlying the identified aspects and clusters of aspects based on a product of loading and score correlation (raised to corr.power). Clusters of aspects driven by the same genes are determined based on the parameter "distance.threshold".

Description

Collapse aspects driven by the same combinations of genes. (Aspects are some pattern across cells e.g. sequencing depth, or PC corresponding to an undesired process such as ribosomal pathway variation.) Examines PC loading vectors underlying the identified aspects and clusters of aspects based on a product of loading and score correlation (raised to corr.power). Clusters of aspects driven by the same genes are determined based on the parameter "distance.threshold".

Usage

pagoda.reduce.loading.redundancy(
  tam,
  pwpca,
  clpca = NULL,
  plot = FALSE,
  cluster.method = "complete",
  distance.threshold = 0.01,
  corr.power = 4,
  abs = TRUE,
  n.cores = 1,
  ...
)

Arguments

tam

output of pagoda.top.aspects(), i.e. a list structure containing the following items: xv: a matrix of normalized aspect patterns (rows: significant aspects, columns: cells) xvw: corresponding weight matrix gw: set of genes driving the significant aspects df: text table with the significance testing results

pwpca

output of pagoda.pathway.wPCA(), i.e. a list of weighted PCA info for each valid gene set

clpca

output of pagoda.gene.clusters() (optional) (default=NULL). The output of pagoda.gene.clusters() is a list structure containing the following fields: clusters: alist of genes in each cluster values xf: extreme value distribution fit for the standardized lambda1 of a randomly generated pattern tci: index of a top cluster in each random iteration cl.goc: weighted PCA info for each real gene cluster varm: standardized lambda1 values for each randomly generated matrix cluster clvlm: a linear model describing dependency of the cluster lambda1 on a Tracy-Widom lambda1 expectation

plot

boolean Whether to plot the resulting clustering (default=FALSE)

cluster.method

string One of the standard clustering methods to be used (default="complete")

distance.threshold

numeric Similarity threshold for grouping interdependent aspects (default=0.01)

corr.power

numeric Power to which the product of loading and score correlation is raised (default=4)

abs

boolean Whether to use absolute correlation (default=TRUE)

n.cores

numeric Number of cores to use during processing (default=1)

...

additional arguments are passed to the pagoda.view.aspects() method during plotting

Value

a list structure analogous to that returned by pagoda.top.aspects(), but with addition of a $cnam element containing a list of aspects summarized by each row of the new (reduced) $xv and $xvw