## Reference database downloads

This directory contains the GOTTCHA2 RefSeq-GTDB multi-domain+ reference databases, FASTA files, and SHA-256 checksums. Choose the database archive appropriate for your profiling mode.

| File or directory                         | Size   | Description |
|-------------------------------------------|--------|-------------|
| FASTA/                                    | -      | Reference sequence files in FASTA format. |
| gottcha_db_fast.tar                       | 32 GB  | Reference database archive for GOTTCHA2 fast profiling mode. |
| gottcha_db_fast.tar.sha256                | -      | SHA-256 checksum |
| gottcha_db_fast_p_Knoellia.tar            | 32 GB  | Fast database plus Knoellia signatures. |
| gottcha_db_fast_p_Knoellia.tar.sha256     | -      | SHA-256 checksum |
| gottcha_db_standard.tar                   | 198 GB | Reference database archive for GOTTCHA2 standard profiling mode. |
| gottcha_db_standard.tar.sha256            | -      | SHA-256 checksum |
| gottcha_db_standard_p_Knoellia.tar        | 164 GB | Standard database plus Knoellia signatures. |
| gottcha_db_standard_p_Knoellia.tar.sha256 | -      | SHA-256 checksum |

Sizes are approximate archive sizes reported by the download server.

### Verify and extract

Download the selected archive and its matching `.sha256` file into the same directory. Verify the download before extracting it. For example:

```bash
sha256sum -c gottcha_db_fast.tar.sha256
tar -xvf gottcha_db_fast.tar
```

Replace the filenames with those of your selected database.

## Genomes Used in Signature Identification

The raw signature sequences are sorted by the taxonomy from superkingdom(domain) to species. Details can be found here: https://poeli-lanl.github.io/gottcha2_stats/RefSeq-GTDB-multi-domain_p/

Bacterial (n = 40,285) and archaeal (n = 616) representatives were drawn from NCBI Refseq assemblies at the "complete genome" or "chromosome" level with taxonomy according to the GTDB r220 [ref]. The genomes of viruses (n = 14,979), protozoa (n = 123) and plastid (n = 14,428) were obtained from NCBI RefSeq. Fungal genomes (n = 2,644) and their taxonomy were sourced from JGI's MycoCosm database [ref]. To reduce host-derived false positives, plant (n = 265) and human (n = 2) genomes were included only during signature identification to flag and remove consensus k-mers shared with these hosts; plant- and human-derived signatures were not stored in the final database. Plant genomes were additionally filtered by masking k-mers shared with plastid genomes prior to signature identification.

** RefSeq-GTDB-multi-domain+ (# genomes) **
  - GTDB r220
      - Bacteria (B)                   40,285
      - Archaea (A)                       616
  - Refseq r223
      - Viruses (V)                    14,979
      - Protozoa (Pz)                     123
      - Plastid genomes (Pt)           14,428
      - remove (x) Plant (Pn)             265 *Plant signatures are not stored
        - plastid filtered (PmaskedPn)
      - remove (x) Human (H)                2 *Human signatures are not stored 
  - JGI MycoCosm
      - Fungal genome JGI (F)           2,644

** plus Knoellia  (# genomes) **
  - origianl RefSeq-GTDB-multi-domain+
  - GTDB r232
      - Knoellia                           18

The Knoellia signatures were generated as an add-on by excluding k-mers shared with any genome in RefSeq-GTDB-multi-domain+. However, these Knoellia genomes were not included when the original RefSeq-GTDB-multi-domain+ signatures were identified. The minimap2 index was built with -k24 -w16 -I7G parameters, so it's smaller than the original one.