Ignore over-occuring kmers?

[tseemann]

Would an option to ignore over-occurring kmers make mash more robust against large repeat families and multi-copy plasmids?

mash estimates the coverage in -r mode, and it uses -m for a min freq, but maybe 2*est_cov would be a good max freq?

eg. -M 2 would ignore kmers with freq > 2*est_cov

[tseemann]

I've just realised Finch does something like this already https://github.com/onecodex/finch-rs