I am wondering if anyone has been successful in creating a custom kraken database that has bacteria, human and fungal genomes.
We have a server that has 128 GB RAM and I was able to download the taxonomy, add the taxa to the library. I started with adding bacteria, fungi, human and viral genomes to the database. While this worked, the kraken-build also worked for 10 hours until it built the ‘database.jdb’ which was 176 GB. Then it stopped giving this error ‘db_sort: unable to mmap database: Cannot allocate memory’. I used the jellyfish-hash 6400M.
So, I tried with only bacterial genomes where it created a ‘database.jdb’ for 138 GB and stopped giving the same error. Next, just to confirm this was not a technical error, I tried by building the database with only fungal and human genomes and this worked with a database of 31 GB. But this doesn’t really serve the whole purpose of having a single database that can detect fungal, bacterial and human.
It would be great to have your inputs or direct me to anyone who has worked with this.