Wrapper function to compute all by all NBLAST scores for a set of neurons
nblast to compute the actual scores. Can accept
either a neuronlist or neuron names as a character vector. This is a thin
wrapper around nblast and its main advantage is the option of "mean"
normalisation for forward and reverse scores, which is the most sensible
input to give to a clustering algorithm as well as the choice of returning
a distance matrix.
nblast_allbyall(x, ...)"nblast_allbyall"(x, smat = NULL, db = getOption("nat.default.neuronlist"), ...)"nblast_allbyall"(x, smat = NULL, distance = FALSE, normalisation = c("raw", "normalised", "mean"), ...)
- Input neurons (neuronlist or character vector)
- Additional arguments for methods or
- the scoring matrix to use (see details of
nblastfor meaning of default
- A neuronlist or a character vector naming one. Defaults to value of options("nat.default.neuronlist")
- logical indicating whether to return distances or scores.
- the type of normalisation procedure that should be
carried out, selected from
'mean'(i.e. the average of normalised scores in both directions). If
distance=TRUEthen this cannot be raw.
nat already provides a function
nhclust for clustering, which is a wrapper for R's
nhclust actually expects raw scores
It would be a good idea in the future to implement a parallel version of this function.
library(nat) kcs20.scoremat=nblast_allbyall(kcs20) kcs20.hclust=nhclust(scoremat=kcs20.scoremat) plot(kcs20.hclust)