vsearch: VSEARCH version 1.0.10

VSEARCH 1.0.10 incorporates a number of fixes and changes:

• fix for bug in alignment in rare cases
• fix for alignment of long sequences (>15000bp)
• fix for lack of detection of under- or overflow in SIMD aligner
• default maxseqlength set to 50000
• new linear memory aligner for alignment of longer sequences reduces memory requirements
• a new cluster_size command that sorts sequences by decreasing abundance before clustering
• cluster_fast will now sort sequences by decreasing length, then by decreasing abundance and finally by sequence identifier
• help messages for clustering have been improved
• in uc files, perfect alignment is indicated with "=" sign
• meaning of userfields qlo, qhi, tlo, thi changed for compatibility with usearch; new userfields qilo, qihi, tilo, tihi gives coordinates ignoring terminal gaps
• a few other small fixes

Alignment of long sequences is slow because it is performed by an exact and optimal algorithm. VSEARCH works best with sequences less than 5000 bp long, but should now work correctly with longer sequences as well.

These changes were originally introduced in version 1.0.8, but there were not official releases for 1.0.8 and 1.0.9.