Slacken is available on GitHub (github.com/JNP-Solution...) and reference libraries are available on S3 thanks to AWS Open Data sponsorship.

Feedback very welcome. I'd be happy to answer any questions or assist people in getting started. We want Slacken to be as accessible as possible.

2) We show that dynamically tailoring a genomic reference library to the samples being classified greatly increases the fraction of species and strain level classifications (making them more specific) as well as improving Bracken quantification.

1) We introduce a new implementation of the Kraken 2 method on Apache Spark, which has comparable cost-performance when classifying multiple samples.

Particularly with a focus on making the software accessible for people with no Spark experience.

I can imagine that people who are entering into software development now might get the false impression that there's only accidental complexity and AI is our only hope to temper it. But you only get to understand simplicity by developing your own taste for it (by fighting complexity for long enough)

One challenge I think young people are facing is that you have to wade through so much accidental complexity before you start seeing the light. It's only in my mid 30's that I think I understood how to value simplicity and elegance. Before that I was not seeing the forest for the trees a lot.

CDC datasets have been saved. But you can still help by seeding.

This resolves an inherent conflict between scalability and precision in Kraken 2.