problem kraken build standard database

Question

[david.christiany]

Hi,

After having successfuly installed kraken and jellyfish, I'm trying to build the standard database. I run the following command on a virtual Ubuntu with 102 Go of RAM and 16 cpu :

kraken-build --standard --threads 16 --db $DBNAME

 

Found jellyfish v1.1.11
Skipping download of bacterial genomes, already downloaded here.
Skipping download of viral genomes, already downloaded here.
Kraken build set to minimize disk writes.
Creating k-mer set (step 1 of 6)...
Found jellyfish v1.1.11
Hash size not specified, using '11637743885'

My problem is that the first step is still running after 2 days and seems to never end. It is suppose to take only 10 minutes with 16 threads according to the manual.

Is there someone who have already had this issue and solve it ?

Thanks in advance.

[himani.khurana1]

It has been almost 3 days that it is at step1 of kraken Creating k-mer set (step 1 of 6) Found jellyfish v1.1.10 Hash size not specified, using 18937420000

The path for database is in external hard disk

Although no error reported but is there anything wrong that I might be doing I have my server with 128 GB memory and 1TB space in hard disk It is already consuming around 107 GB I have also specified threads=20

Is there something else I should do?

[Ram]

Why have you added this as an answer? This belongs as a comment on the top level post (I've moved it to one now). Please do not add answers unless you're answering the top level post.

Answer 1

[david.christiany]

Hi,

I used a more powerful node on the cluster on which I was building the kraken database. I had a big custom database, so that's why I needed more memory, I think it was about 400Gb. Of course it depends on the initial size of your database before you start to build it and on the thread parameter.

If you want to build a big (or even relatively big) kraken custom database, there's no secret parameters, I think you just need more memory.

[Ram]

This should be a reply to himani.khurana1's comment, not an answer.