Metagenomics · Snakemake workflow
Aviary¶
Aviary assembles metagenomes, recovers metagenome-assembled genomes (MAGs), and coordinates their annotation in reproducible local or HPC Snakemake workflows. It supports short-read, long-read and hybrid analyses.
$aviary complete
A complete analysis in one command¶
aviary complete \
-1 sample_R1.fastq.gz \
-2 sample_R2.fastq.gz \
--output sample_aviary \
--max-threads 8 \
--n-cores 8
See the first complete analysis walkthrough for what each stage does and where its outputs land.
The commands¶
Six analysis subcommands cover the workflow. complete runs the metagenome
stages end to end; the others let you enter or leave at a specific point.
Two utility commands, configure and
build, manage reference paths and dependency environments.
complete
reads annotated MAGs
Every stage in one command: assembly, binning, refinement, then annotation. Works from short reads, long reads, or both. Already have an assembly? Pass --assembly and Aviary picks up from binning instead of starting over.
assemble
reads contigs
Step-down hybrid assembly using long and short reads together, playing each to its strengths — long reads for contiguity, short reads for accuracy. Either type alone works too. Several short-read samples with no long reads are co-assembled with MEGAHIT or metaSPAdes.
recover
assembly MAGs
Sorts contigs into metagenome-assembled genomes using several binning algorithms rather than trusting any single one, then refines the result and reports quality and taxonomy per MAG. With no assembly it runs the assembly pipeline first; with several it enables SemiBin2 multi-sample binning.
annotate
MAGs annotations
Adds functional annotations with EggNOG and taxonomic classifications with GTDB-Tk. Assemblies can be passed alongside when targeting QUAST QC. CheckM2 quality assessment belongs to the recover/complete binning path, not the default annotation target.
cluster
many runs representatives
Run enough samples and the same organism turns up repeatedly. Galah collapses those near-duplicates across finished Aviary runs and picks one representative genome per cluster — 97% ANI by default, adjustable with --ani.
isolate
pure culture genome
A long-read-first Flye assembly and polishing path for a single organism from pure culture. Optional short reads add another polishing round. For metagenomic data, use assemble or recover instead.
Find what you need¶
-
How reads become assemblies, candidate genomes and annotated MAGs.
-
End-to-end examples with real commands.
-
Every command, option, default and output.
-
Validation messages, failed rules, resource limits.
Using Aviary in research? See the citation guide for Aviary and its upstream tools. Aviary is released under GPL-3.0.