What proportion of bases in your query were ambiguous? If it was a fairly low percentage (~10-20%?), I think you could probably get away with using LASTZ, treating all ambiguous bases as completely ambiguous (--ambiguous=iupac), then write a short script to go through the alignments afterward and filter out ones that don't have a reference base that matches the IUPAC character in the query.

But that's the lazy approach :) Good article.