Weather could provide a method for comparing multiple samples at once

[PingWang7]

Kraken2 is an excellent tools to align microbial sequences.

While in each run, loading the database is a rather time-consuming process.

I wonder if future developers could provide a way to load the database only once, and align multiple samples at once.

[ChillarAnand]

There is --memory-mapping option for this purpose. You can read the kraken2 manual regarding this.

https://github.com/DerrickWood/kraken2/blob/master/docs/MANUAL.markdown#system-requirements

I have also written a detailed tutorial on --memory-mapping and other tips to improve classification speed.

https://avilpage.com/2024/07/mastering-kraken2-performance-optimisation.html

[PingWang7]

@ChillarAnand Thanks a lot. It's really a good solution to use --memory-mapping option in kraken2 to speed up, however, sometimes it is unclear why using this parameter results in slower runtime compared to not using it.

[ChillarAnand]

It all depends on how many index pages are in RAM already. As mentioned in the article, you can use vmtouch to make sure you get consistent performance.