A pair of identical circular sequences can differ only in their breakpoints. In the example below, the two sequences are identical when treated as circular, but their breakpoints differ:
>seq_a
CGAAACGTTCTT
>seq_d
GTTCTTCGAAAC
As a result, they produce different hashes. To enable vclust deduplicate to detect this scenario (for example, through a --circular option), a lexicographically minimal string rotation algorithm could be used to normalize breakpoint positions in circular sequences. I implemented this approach in seq-hasher here.
A pair of identical circular sequences can differ only in their breakpoints. In the example below, the two sequences are identical when treated as circular, but their breakpoints differ:
As a result, they produce different hashes. To enable
vclust deduplicateto detect this scenario (for example, through a--circularoption), a lexicographically minimal string rotation algorithm could be used to normalize breakpoint positions in circular sequences. I implemented this approach inseq-hasherhere.