Data Products and Interpretation

MICOS-2024 produces more than files. It produces analysis claims in different maturity bands. This page explains how to read the output directories as evidence products rather than opaque folders.

Output families

Output family	Typical location	What it represents	How to read it
Cleaned reads	results/quality_control/	Post-filter sequencing input	Trust boundary for all downstream analysis
Classification reports	results/taxonomic_profiling/	Per-sample taxonomic assignment summaries	Good for composition review and abundance ranking
BIOM tables	results/taxonomic_profiling/feature-table.biom	Structured abundance matrix	Feedstock for diversity workflows
Diversity artifacts	results/diversity_analysis/	Ecological comparisons	Good for cohort separation and richness summaries
Functional tables	results/functional_annotation/	Pathway or feature abundance views	Useful for pathway-level hypotheses
Summary reports	final HTML outputs	Review-friendly synthesis	Best entry point for stakeholders outside the codebase

Reading the results directory

results/
├── quality_control/
├── taxonomic_profiling/
├── diversity_analysis/
├── functional_annotation/
└── micos_summary_report.html

This layout matters because it mirrors the pipeline itself. A reviewer can infer provenance: later directories depend on earlier ones.

What counts as strong evidence

Strongest current signals

• explicit CLI commands in micos/cli.py,
• result folder contracts documented by the CLI reference,
• shell wrapper tests under tests/test_shell_wrappers.py,
• configuration templates that align with those runtime surfaces.

Weaker but still important signals

• advanced helper scripts in scripts/,
• broader template configuration for analyses that are not yet fully unified in the stable CLI,
• container and WDL assets that describe intended execution ecosystems.

Interpreting with caution

Three mistakes are common in metagenomics documentation:

1. treating classification output as if it were contamination-proof,
2. treating diversity plots as self-explanatory,
3. treating every script in the repository as equally stable.

MICOS-2024 is more credible when those distinctions are explicit.

Suggested review path for evaluators

If you are evaluating technical maturity:

1. inspect taxonomic_profiling/ outputs first,
2. verify how they feed feature-table.biom,
3. then assess whether diversity and functional outputs are explained with enough provenance.

Suggested review path for users

If you are operating the platform:

1. confirm QC outputs look reasonable,
2. check taxonomy reports for expected sample signal,
3. only then invest time in diversity or functional interpretation.