ISMB 2026 Workshop: k-mer Methods and Sketching

This tutorial covers sourmash and YACHT for k-mer-based analysis of genomic and metagenomic data. It is designed for a one-hour session but includes additional material for self-directed exploration.

The two tutorials share a single dataset and tell one story. We start with a metagenomic sample and a handful of candidate reference genomes, and we work our way from raw sequence to a statistically supported answer to the question which of these genomes are actually in my sample?

• Sourmash tutorial: what a sketch is, how to build and compare sketches, and why ANI is a better similarity measure than raw containment.
• YACHT tutorial: turning those same sketches into a hypothesis test for genome presence/absence, and how that differs from taxonomic profiling.

---

Prerequisites

conda or miniconda must be installed. See the miniconda installation guide if you need to set this up.

---

Setup

1. Create the conda environment

conda create -n ISMBtutorial -c bioconda -c conda-forge yacht=1.3.2
conda activate ISMBtutorial

yacht depends on sourmash, so this single environment covers both tutorials.

2. Create a working directory and download the data

Everything below runs from one working directory. Create it, download the demo dataset into it, and make a folder to hold the sketches we will build:

mkdir ISMBtutorial && cd ISMBtutorial
yacht download demo --outfolder ./demo
mkdir sketches

This gives you:

ISMBtutorial/
|-- demo/
|   |-- query_data/query_data.fq      # a metagenomic sample (reads)
|   |-- ref_genomes/                   # 15 candidate reference genomes (GTDB)
|   |-- toy_genome_to_taxid.tsv        # maps each genome to a taxonomy ID
|-- sketches/                          # (empty; we will fill this in)

The demo dataset is small and self-contained. The query_data.fq sample was simulated from 5 of the reference genomes at 0.5x coverage, so we have a known ground truth to check our answers against. The 15 reference genomes are randomly selected GTDB representatives.

All commands in both tutorials assume you are in the top-level ISMBtutorial/ directory.

---

Tutorials

Work through the tutorials in order:

1. Sourmash tutorial
2. YACHT tutorial

---

Optional: extra data for self-study

The classic sourmash tutorials use an E. coli dataset (reads, an assembled genome, and a collection of 50 strain signatures) and a pair of small synthetic genomes. These are not needed for this workshop, but if you want to work through the upstream sourmash tutorial on your own afterwards, you can grab them with:

# (optional, self-study only) run from inside ISMBtutorial/
mkdir -p extras && cd extras
curl -L https://osf.io/ruanf/download -o ecoliMG1655.fa.gz
curl -L https://osf.io/q472x/download -o ecoli_ref-5m.fastq.gz
curl -O -L https://github.com/sourmash-bio/sourmash/raw/latest/data/eschericia-sigs.tar.gz
tar xzf eschericia-sigs.tar.gz && rm eschericia-sigs.tar.gz
cd ..