Provide precomputed skani sketches for GTDB representative genomes

[fplazaonate]

Hi,

I was wondering whether you have considered providing precomputed skani sketches for GTDB representative genomes as part of the GTDB-Tk reference data available for download.

I believe this could offer several advantages:

• It would remove the need to compute skani sketches during the first execution of GTDB-Tk.
• Users would only need to specify the GTDB-Tk data path (GTDBTK_DATA_PATH), without also having to provide a separate path to the skani sketches.
• It may make the inclusion of representative genome FASTA files in the GTDB-Tk reference package unnecessary (or significantly reduced), potentially decreasing the package size and speeding up downloads.

Thanks again for all your work,
Florian

[pchaumeil]

Hi Florian,

Thanks for the message.
This transition has actually been discussed recently with the size of R232 getting quite large.

We are planning to do exactly that for the release of R232, with a fixed version of skani.

Cheers,
Pierre

[jolespin]

Would the genome assembly fasta files still be needed for pplacer or were these just used for ani searches?

[fplazaonate]

Pplacer uses only marker genes, not whole genomes.

[jolespin]

Nice, so this would dramatically decrease the amount of files in the database making it a lot quicker to decompress.