June 30, 2010
* First stand alone release

March 7, 2011
* 2.0.1 release
* Maximum number of input sequences was increased to 15999.
* Improvements to reduce time for pre-alignment sequence clustering.
* Added limit on number of domain hits per input sequence that can
  be set with the -num_domain_hits option.

June 13, 2013
2.0.2 release
New features:
* Align two multiple sequence alignments
* Use pre-computed CDD search results in BLAST archive format
Improvements:
* BLAST+ 2.2.28
Bug fixes:
* Nexus alignment format
* Segmentation fault on out-of-memory
* Alignment problems with clustering input sequences

January 31, 2017
2.1.0 release
Improvements:
* More memory efficient sequence clustering
* No longer prints warning message about all sequences in one cluster
Bug fixes:
* Fixed clustering results for sequences with ambiguous amino acids

August 4, 2023
3.0.0 release
New features:
* Usage reporting - Help improve COBALT by sharing limited information about your executions. Details on the information collected, how it is used, and how to opt-out can be found here: https://www.ncbi.nlm.nih.gov/books/NBK563686/. To opt out set the environment variable BLAST_USAGE_REPORT=false.
Improvements:
* More accurate alignments
* Updated Conserved Domain Database (CDD), based on CDD release version 3.20
* New CDD with adaptive word threshold, which results in faster RPSBLAST searches and multiple alignment computation.
* Reads CDD in version 5 format (BLASTDBv5), more about BLAST database version 5 in https://ftp.ncbi.nlm.nih.gov/blast/db/v5/blastdbv5.pdf
* Query clustering (the -cluster option) is turned off by default
* The default CDD search E-value cutoff (the -rps_evalue option) is now 0.01
Bug fixes:
* Fixed memory leaks
* COBALT now shows available scoring matrices when an incorrect one is requested
