Tool: Blini: Lightweight nucleotide sequence search and dereplication
Hi, I'm happy to share Blini, a new tool for sequence lookup and dereplication at massive scales, where traditional tools like BLAST or locally-run software might hit resource limits.
What is Blini? Blini is an algorithm for quick nucleotide sequence lookup and dereplication. It combines several k-mer based techniques to estimate average nucleotide identity (ANI) or containment. It is particularly useful for cleaning and characterizing large collections of metagenome-assembled genomes (MAGs).
Key Features:
- Blini is delivered as a single runnable binary with no external dependencies, just grab and run.
- Easy to use; reasonable defaults and minimal options for configuration.
- Quick and lightweight; clustering a 570MB viral dataset with 19K genomes takes 11 seconds and uses 80MB of RAM; searching a 10GB bacterial reference for 100K queries, 10KB each, takes 26 seconds and uses 2GB of RAM. All using a single thread.
- Adjustable resolution; change the "scale" parameter to balance resource consumption vs effectiveness on short queries.
If you try it, I'd love to get your feedback!
- Read here: https://doi.org/10.21105/joss.09494
- Download here: https://github.com/fluhus/blini/releases
- Docs here: https://github.com/fluhus/blini
- Try it here, with a microbial reference: http://64.227.104.66
(your browser may complain that it's not secure, that's because the site doesn't have its own address yet)
• 764 views
•
link
updated
by
Istvan Albert
•
written
by
Amit Lavon •
•
written
by
Amit Lavon •
0 answers
No answers yet.
Log in to answer this question.
More posts like this
-
Tool: Quickly Build Your Own GEOmetadb with geokit
written by Yun •
-
Tutorial: How to align RNA and DNA 3D structures using ARTEMIS
written by Eugene F. Baulin •
-
Tool: NOEL: An extremely fast Non-Overlapping Exon Length calculator written in Rust
written by alejandrogzi •
-
Tool: Genotify: Fast, lightweight gene lookup and summarization
written by jared.andrews07
One suggestion for your READMEe is to provide a realistic example. Say take the representative viral genome from NCBI, those reasonably small and demonstrate the tool via those examples.
Right now your examples are input.fasta and output.csv and the reader has no idea what is inside these files, what does the tool do.
Similarly take 100 SARS COV 2 genomes and cluster them.
Basically I would like to know, without running the tool, what the outputs look like, what it results look like.
Thank you for your feedback. I've added explanations of the output formats to the README, as well as some real-world examples. Hope this helps!
Very cool examples, and just reiterate, move these examples first and one should not need to click to expand information
The only and sole reason for using any tool is the output, so why hide the most important information?
Then, the reference file that you use in your example is 43 GB compressed, that blini can handle it is super impressive! But how long does that process take 1 hr, 10hrs, 100hrs? You should indicate that, sketching takes X amount of time and the query now takes Y amount of time.
I am trying to redo the example, but looks like it will take a while. It would be so much more helpful if I approximately knew beforehand what to expect.
Maybe there could be another smaller file that we can download and run in 5 minutes that demonstrate the general usage. Then show the very impressive gigantic search as well.
The tool looks amazing, but it doesn't communicate clearly or make a strong case for what it does. I know it feels unfair, but that's reality. If people don't understand what something does in a short amount of time, they move on, and a good product gets set back.
And in the README should contrast it to a typical usage, and why this new alternative is better.
The README is the your main entrypoint to people's mind, not the paper. We read the paper after you convinced us in the README that it is worth the time. Time/attention is the greatest constraint nowadays.