Kütral: the Di Genoma Lab compute platform

Kütral means “fire” in Mapudungun. It is the high-performance computing platform used by Di Genoma Lab to process large sequencing datasets, run production workflows, and prototype new algorithms for genomics.

Kütral currently provides 1000 CPU cores, 7 TB of RAM, and 200 TB of storage, connected through 40 Gbit/s InfiniBand. This infrastructure supports demanding workloads such as human genome assembly, somatic structural variant calling, long-read analysis, population genomics, metagenomics, and large-scale workflow execution.

Beyond total capacity, the cluster includes high-memory nodes for assembly and other memory-intensive tasks, as well as GPU-enabled resources for machine learning and accelerated analysis. Most of the platform is dedicated to genomics and computational biology workloads generated by sequencing and software development projects in the lab.

Software environment

Kütral is managed as a reproducible research platform rather than an ad hoc compute pool. The environment includes:

• BeeGFS for high-performance shared storage
• Slurm for job scheduling and resource management
• Ansible for automated configuration and maintenance
• Nextflow for scalable, reproducible workflow execution
• Open OnDemand for browser-based access to the cluster

This stack allows us to move efficiently from method development to stable production runs while keeping analyses reproducible and auditable.

Why it matters for our research

Local compute is essential for the work we do. It allows us to:

• iterate quickly on new assembly and variant-calling methods
• process sensitive clinical or institutional datasets under local control
• optimize workflows for high-throughput long-read and cancer genomics analyses
• train researchers and engineers in realistic, production-grade computational environments

Kütral is a core part of how Di Genoma Lab connects sequencing, software, and biological interpretation.