kraken2-build not finish

Question

[spatroca]

Hello!

I am trying using Kraken2 to verify viruses in a virome rawdata. For this, I need the nr (NCBI non-redundant protein database).

I did the download of the taxonomy using:

kraken2-build --download-taxonomy --protein --threads 24 --use-ftp --db /data/02_db/kraken2/library/nr/

I did the download of the library using:

kraken2-build --download-library nr --no-masking --protein --threads 24 --use-ftp --db /data/02_db/kraken2/library/nr/

And build the library using:

kraken2-build --build --protein --threads 80 --db /data/02_db/kraken2/library/nr/

However, five days after the beginning of this step, the library-build didn't show any error and didn't finish, I killed the run because I am not sure if will finish.

Please, anyone could tell me if have anything wrong with these commands that I used? Or do I need to include more parameters?

[Mensur Dlakic]

I haven't done this in a while because I don't use kraken any more, but my recollection is that this process is both memory- and disk-intensive. I don't think doing it with 80 threads will do you any good, and in fact it may hurt the performance if the memory is not shared between the threads.

The thing I always do with these long processes is to check the computer load, and how much swapping is going on (sometimes this shows up as many kworker processes). If the load is low and the swapping is high, it usually indicates that either the RAM or disk speed are main problems - sometimes both.

By the way, next time maybe first ask the question here before killing the program.

[Mensur Dlakic]

May be worth trying with --fast-build:

https://github.com/DerrickWood/kraken2/issues/492

[spatroca]

Hi, thank you so much for your return. I tested include the parameter --fast-build, and worked. I apreciate your help.

[spatroca]

Hi, thank you so much for your return. You said that don't use kraken anymore. Are you using another program in place of kraken?