Deployment Modes

MICOS-2024 can be deployed in several ways, but they are not equivalent. This page explains what each mode is trying to optimize.

Mode 1, direct Python execution

Best for:

local development,
debugging,
repositories where the Python environment is already controlled.

Strengths:

shortest feedback loop,
easiest way to inspect CLI behavior,
direct alignment with micos/cli.py.

Mode 2, shell-wrapper execution

Best for:

backwards-compatible automation,
teams migrating from older script-first workflows.

Important detail:

The wrapper scripts are now thin delegators, not independent pipeline engines. This is good. It reduces logic drift.

Mode 3, workflow-driven execution

Best for:

larger execution ecosystems,
workflow-engine integration,
explicit step-level orchestration.

Relevant assets:

steps/01_quality_control/fastqc.wdl
steps/02_read_cleaning/kneaddata.wdl
steps/03_taxonomic_profiling_kraken/kraken2.wdl
steps/04_taxonomic_conversion_biom/kraken-biom.wdl
steps/05_taxonomic_visualization_krona/krona.wdl
steps/06_qiime2_analysis/*.wdl

Mode 4, container-backed execution

Best for:

reproducible dependency stacks,
collaborator handoff,
isolating environment drift.

Relevant assets:

deploy/docker-compose.example.yml
containers/singularity/*.def

Deployment matrix

Need	Preferred mode
Fastest local iteration	Direct Python CLI
Compatibility with legacy shell workflows	Shell wrappers
Portable step graph	WDL assets
Reproducible environment package	Containers

Practical recommendation

For most contributors:

operate the stable CLI first,
use workflow assets when integrating with orchestration systems,
use containers to freeze environments,
use wrappers only when you need the legacy ergonomics.