Closed recursion-ninja closed 5 years ago
This was mostly implemented via the C FFI binding to the C code inherited from POY. I wrote a wrapper on the Haskell side and refactored some of the Haskell FFI code to separate the dense TCM generation and queries from the C DO alignment code.
This has been implemented and integrated into the dynamic character metadata.
We should expand the C FFI to allocate a 3D "TCM" which calculates the cost between any 3 ambiguous alphabet states of size eight or fewer. This should improve the efficiency over the general case with the memoized TCM and might be able to be integrated into the 3D (three sequence) C alignment code.