healthyPlant / PhytoPipe

10 stars 1 forks source link

"kraken2 build" requires more than 700GB memory, how can I do? #6

Open xhu556 opened 10 months ago

xhu556 commented 10 months ago

While I was executing the "update database" script I received a message at the "kraken2 build" step saying that more than 700GB was required to build the NCBI nt database.

xhu556 commented 10 months ago

If you run "kraken2-build", please add "--max-db-size 128000000000" to limit the kraken database size to 128Gb, or to any size you want, such as “256000000000” (256Gb), or “64000000000” (64Gb).

kraken2-build --build --threads 32 --db /path/to/my/database/kraken_db --max-db-size 128000000000

If you run the script updataDatbase.sh, please download the updated version. The database is limited to 128Gb in the script. You can change it to any size you want by editing "--max-db-size 128000000000" in the line 92.

We checked how Kraken2 database size affects the classification results (https://github.com/healthyPlant/PhytoPipe/wiki#dbsize). Our conclusion is the large database is better than the small one.