• blast each feature against the databases, store in a map from feature name to a list of matches with that feature
• traverse the ordered list of features and find the minimum cost/length combinations of fragments/vectors to assemble a vector with all the features in order (returning all solutions)
• add dbs flag to 'defrag features find' allow checking of features in databases as well will add 'defrag ls fragment [name]' instead
• return pareto optimal solutions

Feature CLI Grammar

[name]:fwd/rev:l#:r#

• fwd/rev = whether it's in fwd or rev direction, fwd assumed
• l# = upper bound on "padding" in bp to consume from the left matched fragment
• r# = upper bound on "padding" in bp to consume from the right matched fragment

default l/r is unlimited for matches that contain adjacent features and 0 for matches without any other additional features

Longterm

• read in Genbank file features and use
• read in multi-FASTA file features
• gap penalty for inserting a new feature? Post assembly, blast each vector solution against the features db and penalize those that added new features