Routines that create identity-by-descent (ibd) coefficients
often output their results as a list of values (i, j, x[i,j]),
with unlisted values of the x matrix assumed to be zero.
This routine recasts such a list into
bdsmatrix form.
Usage
bdsmatrix.ibd(id1, id2, x, idmap, diagonal)
Arguments
id1
row identifier for the value, in the final matrix.
Optionally, id1 can be a 3 column
matrix or data.frame, in which case it is assumed to contain
the first 3 arguments, in order.
id2
column identifier for the value, in the final matrix.
x
the value to place in the matrix
idmap
a two column matrix or data frame.
Sometimes routines create output with integer values for
id1 and id2,
and then this argument is the mapping from this internal label to
the “real” name)
diagonal
If diagonal elements are not preserved in the list, this value
will be used for the diagonal of the result.
If the argument appears, then the output matrix will contain
an entry for each value in idlist.
Otherwise only those with an explicit entry appear.
Details
The routine first checks for non-symmetric or otherwise inconsistent input.
It then groups observations together into ‘families’ of
related subjects, which determines the structure of the final matrix.
As with the makekinship function,
singletons with no relationships are first in the output matrix, and
then families appear one by one.
Value
a bdsmatrix object representing a
block-diagonal sparse matrix.