How do I create consensusMatrix equivalent with respect to codons? That is, how do I count codon frequencies by sequence site to generate something similar to what consensusMatrix returns for a DNAStringSet or a AAStringSet object? I have tried letterFrequency but it generates frequencies per. sequence not per site.

dna1 <- DNAStringSet(c("TTGATATGGCCCTTATAA", "CTCAGATCACTCTTTGGC"))
consensusMatrix(dna1, baseOnly=TRUE)

The sample code above returns a 5 by 18 matrix with respect to DNA bases (and an "other" category). How do I generate a 65 by 6 equivalent in terms of codons?