Below is a summary of the key changes in this pull request:
Accepts multiple alignments in one run, with multiple corresponding outputs (adding a required argument -o to make this possible).
Adds utilities for locality-sensitive hashing (LSH) and approximate near neighbor lookup.
Adds option (--id, together with --id-m and --id-frac) to perform differential identification -- i.e., designing guides for each alignment that are specific to it.
Clusters potential guide sequences before taking a consensus: rather than approximating an optimal guide sequence to cover a set S of sequences by taking the consensus of a guide region over S, this now uses LSH to produce a rough clustering of the guide regions in S and then produces a consensus over just the largest cluster.
Below is a summary of the key changes in this pull request:
-o
to make this possible).--id
, together with--id-m
and--id-frac
) to perform differential identification -- i.e., designing guides for each alignment that are specific to it.