Fig. 1

Multipartite Choice Function. The choice function aims to find optimal variants from a set of protein sequences. a) Sequences are aligned; b) A consensus alignment is produced: on a column-by-column basis the choice of amino acid for each sequence that optimises the alignment score for that column is chosen as a representative; c) A binary representation is produced from the original alignment: for each position in alignment, a 1 is assigned if the amino acid matches the consensus, and a 0 is assigned if it does not. This leaves a sequence of vertical binary strings. The aim is to find a single vertical binary string that agrees with (i.e. is a bitwise subset of) as many as possible of these, and that is also compatible with the category constraints. The best such string in this case is shown to the right in green. d) The result