Sparse 0/1 aggregation matrix mapping ids to groups

Construct a sparse matrix with one row per id and one column per group, containing a 1 where an id belongs to a group. Right-multiplying a connectivity matrix by grouping_matrix(colnames(M), group) sums the columns of M within each group (e.g. collapsing partner neurons to cell types).

Usage

grouping_matrix(ids, group)

Arguments

ids

Character vector (or coercible) of identifiers, typically the row or column names of a matrix to be aggregated.

group

Grouping labels. Either a vector parallel to ids or a named vector that will be looked up by ids. Ids whose group is NA are dropped (contribute no 1s), so they vanish from any aggregation.

Value

A sparse Matrix of dimension length(ids) x number of groups with ids as row names and group labels as column names.

See also

effective_connectivity

Examples

library(Matrix)
M <- Matrix(c(1,0,2, 0,3,0), nrow=2, byrow=TRUE,
  dimnames=list(c('a','b'), c('x','y','z')))
# collapse columns x,y -> g1 and z -> g2
G <- grouping_matrix(colnames(M), c(x='g1', y='g1', z='g2'))
M %*% G
#> 2 x 2 Matrix of class "dgeMatrix"
#>   g1 g2
#> a  1  2
#> b  3  0