editspace

Library that implements a Levenstein trie to calculate distance between sequences.
Provides two binaries.

correct takes two arguments,
a maximum edit distance and
a list of known sequences.
It then corrects sequences on the first field of stdin
to the closest sequence in the list of known sequences
(anything in subsequent fields is simply echoed).
If a match can't be found, or two sequences in the list are equally close,
correct prints the corresponding line to stderr
with an additional field explaining the reason for exclusion.
The list of known sequences can have two columns,
in which case, the word on the first column
is replaced by the word on the second column.

cluster takes three arguments,
a maximum edit distance,
a minimum count required to consider a sequence a viable cluster, and
an expected maximum fraction for a sequence to be considered an error.
It counts sequences on stdin,
then runs through them again,
in order of highest to lowest count,
and prints only the sequences that are at least
the maximum edit distance different from any sequences with more than
the maximum fraction expected for error sequences.
Any rejected sequences are printed,
with their counts,
and a reason for exclusion,
on stderr.