Our reference databases may be downloaded.

REFV6

• FASTA (4 MB)
  All refv6 sequences in fasta format. The definition line is a concatenation (using "|") of the vref_id, ref16s_id, accession number, species from silva, the final taxonomic assignment and the source of that taxonomy. When blasting we tend to run use only one copy of the unique v6 region (vref_id = ref16s_id) so as to minimize the redundancy, and later we expand using the ref16s_ids to determine all possible sources of a unique v6 region.

• SQL Data File (4.5 MB)
  A complete SQL dump of our RefV6 database table. This is excised in silico from high quality sequences of the Ref16S table, including the final taxonomy, the source ref16s_id (previously alt_local_gi), and the domain. The ref16s_id is unique. The vref_id (previously local_gi) is non-unique and identifies distinct v6 region tags. All references that have the same variable region sequence are given the same vref_id.
  

REFV3

• FASTA (7 MB)
  All refv sequences in fasta format. The definition line is a concatenation (using "|") of the vref_id, ref16s_id, accession number, species from silva, the final taxonomic assignment and the source of that taxonomy. When blasting we tend to run use only one copy of the unique v3 region (vref_id = ref16s_id) so as to minimize the redundancy, and later we expand using the ref16s_ids to determine all possible sources of a unique v3 region.

• SQL Data File (7.5 MB)
  A complete SQL dump of our RefV3 database table. This is excised in silico from high quality sequences of the Ref16S table, including the final taxonomy, the source ref16s_id (previously alt_local_gi), and the domain. The ref16s_id is unique. The vref_id (previously local_gi) is non-unique and identifies distinct v3 region tags. All references that have the same variable region sequence are given the same vref_id.
  

REF16S

• FASTA (78 MB)
  An aligned fasta of all high quality ref16s sequences (ARB alignment). The definition line is the same as for the refv6 fasta, but it does not include the vref_id.

• SQL Data File (82 MB)
  A complete SQL dump of our Ref16S database table. The original source of this data is SILVA rRNA database project (version 93). Low quality sequences are flagged as deleted (pintail score <=40, sequence quality <= 50 or alignment quality <= 50). All sequences are assigned taxonomy via the RDP (fields: rdp_taxonomy, rdp_boot). Additional taxonomic information is included as it becomes available (fields: other_taxonomy, other_source). A final taxonomic assignment is made for all high quality sequences (fields: taxonomy, taxon_source). Using other_taxonomy preferentially and then RDP for all other sequences. Only RDP classifications with a boot score > 80 are included in the final taxonomy.