Introduction

BioNodulo turns bioinformatics pipelines into visual graphs. Rather than stitching together shell scripts, Snakemake rules, or Nextflow processes by hand, you drag nodes onto a canvas, connect them with edges, set parameters in a panel, and press Run.

The mental model

A BioNodulo workflow is a directed acyclic graph (DAG):

• Nodes wrap a single bioinformatics tool or operation (e.g. FastQC, BWA-MEM, GATK HaplotypeCaller). Each node declares typed input and output ports and a set of parameters.
• Edges carry typed data from an output port of one node to an input port of another. The editor refuses connections between incompatible types — you cannot feed a BAM file into a node expecting a FASTQ.
• A run is one execution of the graph. The engine resolves the topological order, schedules each node, streams logs and progress, and stores outputs.

[FASTQ Input] → [FastQC] → [MultiQC report]
       │
       └──────→ [BWA-MEM] → [Sort/Index] → [GATK HaplotypeCaller] → [VCF Output]

What BioNodulo is good at

• Genomics & variant calling — WGS/WES alignment and variant pipelines.
• RNA-seq & single-cell — quantification, QC, clustering with Scanpy.
• Assembly — de novo genome and transcriptome assembly.
• Long-running, memory-hungry jobs — the cloud runtime checkpoints every step with content-addressed caching so multi-hour and multi-day jobs survive interruptions.

Desktop vs. Cloud

You can move freely between the two: workflows are portable JSON, so a graph you prototype on the desktop runs unchanged in the cloud.

Next steps

• Install the desktop app
• Start in the cloud
• Build your first workflow
• MCP server — drive BioNodulo from Claude, Codex, and other AI clients