v2.2.3: Some issues resolved

• Dependencies added as submodules,
• Fixed rare bug in single linkage when analyzing sequences with zero similarity.
• Deployment scripts added.

v2.2.2

Changes since the previous release:

• 2.2.2 (2022-10-09):
  • Fixed slowdown caused by the duplicate removal (feature added in 2.1.0).
• 2.2.1 (2022-10-05):
  • Pairwise identity (-pid switch) properly calculated as the number of matching residues divided by the shorter sequence length.
• 2.2.0 (2022-10-05):
  • Added possibility to align two pre-aligned profiles (two input files specified).
• 2.1.3 (2022-09-30):
  • Fixed incorrect handling of single sequence sets (first residue cut) or sets containing only duplicates (hang).

v2.1.2

Changes since the previous release:

• 2.1.2 (2022-08-04):
  • Makefile updates improving cross-platform compilation.
• 2.1.1 (2022-08-01):
  • Preserving non-standard amino acid symbols in the output alignment (instead of replacing with X).
• 2.1.0 (2022-07-26):
  • Duplicated sequences are removed prior to the alignment and restored afterwards.
• 2.0.4 (2022-07-25)
  • Fixed stack overflow exception when saving/loading large Newick trees (boost-based algorithm replaced with own non-recursive approach).
• 2.0.3 (2022-05-27):
  • The ordering of the input sequences preserved in the final alignment.
• 2.0.2 (2022-05-24):
  • Alignment allowed as an input (gaps are removed).

v2.0.1

Several fixes to integrate FAMSA 2 with bioconda. If you need precompiled binaries, please take a look at the previous release.

v2.0.0-rc

The following important changes have been introduced:

• Fast guide tree heuristic called Medoid Tree (-medoidtree switch) for ultra-scale alignments:
  • the entire Pfam-A v33.1 in its largest NCBI variant (over 18 thousand families, 60 GB of raw FASTA files) was analyzed in 8 hours,
  • the famility PF00005 of 3 million ABC transporters was aligned in 5 minutes and 24 GB of RAM.
• Remarkable time and memory optimizations - SLINK has been replaced with Prim’s minimum spanning tree algorithm when constructing default (single linkage) guide trees. NOTE: This may change quality results slightly compared to FAMSA 1 due to different ties resolution.
• Neighbour joining guide trees (-gt nj option). NOTE: Neighbour joining trees are calculated with a use of original O(N³) algorithm, thus their applicability on large sets is limited (unless they are used as subtrees with Medoid Tree heuristic),
• Option for compressing output aligment to gzip (-gz switch).
• Compatibility with ARM64 8 architecture (including Apple M1).

v1.6.2

macOS binary added (executables for Windows and Linux are distributed with the previous release).

v1.6.1

• Removed VCL and ASMLIB dependencies.
• Optimizations in the guide tree construction (a few percent speed improvement).

v1.5.12

Changes:

• improved interface for tree import/export (e.g., possibility to export a tree without doing an alignment),
• GPU mode no longer supported.

Bug-fixes:

• fixed bug with -dist_export mode (sequences not ordered as in input FASTA file)

v1.3.2

• Added export of distance matrix.
• Support single input FASTA files (rewrite to output without changes).
• Fix * char emitted for unknown residues, emits X instead.

Guide tree export

v1.2.5

Added export of a guide tree to the Newick format (#11).