Report: Fitness Modules x Pangenome Conservation

Key Findings

Module Genes Are More Core Than Average

• Module genes: 86.0% core vs all genes: 81.5% (+4.5 percentage points)
• Genes assigned to ICA modules are co-regulated functional units, and they skew toward the conserved core genome

(Notebook: 01_module_conservation.ipynb)

Most Modules Are Core

Of 974 modules with >=3 mapped genes:

• 577 (59%) are core modules (>90% core genes)
• 349 (36%) are mixed modules (50-90% core)
• 48 (5%) are accessory modules (<50% core)

The median module is 93.4% core. Most co-regulated fitness response units are embedded in the conserved genome.

(Notebook: 01_module_conservation.ipynb)

Family Breadth Does NOT Predict Conservation

Surprisingly, module families spanning more organisms do NOT have higher core fractions (Spearman rho=-0.01, p=0.914). Families are nearly all core regardless of how many organisms they span. The core genome baseline is so high (~82%) that there’s little room for a gradient.

(Notebook: 02_family_conservation.ipynb)

Accessory Module Families Exist

38 families have <50% core genes — these are co-regulated accessory gene modules conserved across organisms. They may represent horizontally transferred functional units or niche-specific operons.

Essential Genes Are Absent from Modules

0 essential genes appear in any module, confirming that ICA modules only capture genes with measurable fitness variation (non-essential genes with transposon insertions). Essential genes are invisible to ICA because they have no fitness data.

Interpretation

ICA fitness modules are enriched in core genes (86% vs 81.5% baseline, OR=1.46, p=1.6e-87), confirming that co-regulated functional units preferentially reside in the conserved genome. However, the enrichment is modest due to a ceiling effect — the baseline core rate is already very high. The surprising null result for family breadth vs conservation (rho=-0.01) suggests that conservation is a property of individual genes, not of the cross-organism scope of their regulatory module.

Literature Context

• Price et al. (2018) generated the Fitness Browser data and developed the ICA decomposition approach used to identify co-regulated fitness modules. Our analysis shows these modules preferentially reside in the conserved core genome.
• Saelens et al. (2018) demonstrated that ICA outperforms other decomposition methods for microbial gene expression data, supporting the biological relevance of the modules analyzed here.
• Vernikos et al. (2015) reviewed core and accessory genome definitions in bacterial pangenomics. Our finding that 59% of fitness modules are >90% core extends the concept of “core” from individual genes to functionally coherent regulatory units.

Limitations

• Ceiling effect: The baseline core rate is already ~81.5%, limiting the maximum observable enrichment. The +4.5pp difference to 86% is statistically significant but represents a modest absolute effect.
• 29/32 organism subset: 3 module organisms (Cola, Kang, SB2B) lack pangenome links because their species had too few genomes in GTDB for pangenome construction.
• Module membership threshold: The upstream ICA module membership uses |Pearson r| >= 0.3 with max 50 genes per module. This threshold affects which genes are “in” a module and could influence conservation composition.
• Classification thresholds are arbitrary: The 90% and 50% cutoffs for core/mixed/accessory module classification are convenient but not biologically motivated.
• Essential genes excluded: ICA requires fitness data, so essential genes (no transposon insertions) are absent from all modules. This means modules only capture the non-essential portion of the genome.

Data

Sources

Generated Data

References

• Price MN et al. (2018). “Mutant phenotypes for thousands of bacterial genes of unknown function.” Nature 557:503-509. PMID: 29769716
• Parks DH et al. (2022). “GTDB: an ongoing census of bacterial and archaeal diversity through a phylogenetically consistent, rank normalized and complete genome-based taxonomy.” Nucleic Acids Res 50:D199-D207. PMID: 34520557
• Saelens W et al. (2018). “A comprehensive evaluation of module detection methods for gene expression data.” Nat Commun 9:1090. PMID: 29545622
• Vernikos G et al. (2015). “Ten years of pan-genome analyses.” Curr Opin Microbiol 23:148-154. PMID: 25483351

Supporting Evidence

Revision History

• v1 (2026-02): Migrated from README.md
• v2 (2026-02): Added inline figures, notebook provenance, Data section, Literature Context, References