How soedinglab/magmax works
Main components
• CLI Front‐end (main.rs + argument parser)
• Orchestrator / Controller (main.rs)
• Modules (each in src/*.rs):
– Merge Module (merge.rs)
– ReadFetcher Module (readfetch.rs)
– Reassemble Module (reassemble.rs)
– Quality Assessment Module (assess.rs)
– GFA Parser Module (gfaparser.rs)
– Utility library (utility.rs)
• External Tools (invoked at runtime):
– CheckM2 (quality_report.tsv)
– Skani / aligner2counts (for mapid generation)
– Seqtk (for read fetching)
– SPAdes or MEGAHIT (for reassembly)
• Data Stores (directories / files):
– Input Bins Directory (–bindir)
– MapID Directory (–mapdir)
– Reads Directory (–readdir)
– Optional Quality Report File (–qual)
– Output Bins & Reports
Data flow and relationships
• The Orchestrator invokes each module in sequence (or parallel where threaded):
– (Optional) Split Module if “--split” is set (within merge.rs).
– Merge Module: clusters bins by ANI/purity/completeness.
– ReadFetcher: uses mapid files + seqtk to extract relevant reads.
– Reassemble: calls SPAdes/MEGAHIT with fetched reads.
– GFA Parser (if GFA assemblies need graph‐based parsing).
– Assess: runs CheckM2 to produce or consume a quality report.
Architectural patterns
• Modular pipeline pattern: each stage is a self‐contained Rust module.
• Shell‐out pattern: the Reassemble and Assess modules spawn external processes.
• Threaded execution: parallelism controlled by “–threads” argument.
• Functional decomposition: utility.rs provides shared helpers.
Incorporate technologies & frameworks
• Rust & Cargo (runtime & build)
• Conda environment (environment.yml)
• GitHub Actions (release workflow in .github/workflows/release.yml)
Final checklist
[ ] Components labeled: CLI, Orchestrator, Merge, ReadFetcher, Reassemble, GFAParser, Assess, Utility
[ ] Data stores labeled: binsdir, mapdir, readdir, quality_report, output
[ ] External tools: CheckM2, skani/aligner2counts, seqtk, SPAdes, MEGAHIT