Does `is_core` Mean ≥95% Prevalence or Majority Presence?

The corpus does not describe the pangenome “core” gene flag consistently. The pangenome tables and the antimicrobial-resistance fitness-cost analysis describe core as a ≥95% prevalence threshold _1,2, while the structural-annotation analysis describes is_core as presence in the majority of genomes within each species clade ₃. The evidence does not show whether these are two different thresholds or two descriptions of one flag. This matters because statements about gene conservation (whether a gene belongs to the core shared by nearly all genomes of a species or to the variable accessory set) are only comparable across projects if they use the same boundary. That boundary already sits at the centre of pangenome-core-boundary-and-clade-size-bias.

Evidence Sides

Side A: core means ≥95% prevalence The pangenome tables and the antimicrobial-resistance (AMR) fitness-cost analysis both use a ≥95% prevalence threshold for core genes. _1,2 Under this reading, a gene cluster is core only if it is present in at least that share of a species’ genomes.

Side B: core means majority presence The structural-annotation analysis describes is_core as presence in the majority of genomes within each species clade. ₃ Read literally, this is a looser criterion than Side A’s threshold.

Unresolved status The evidence does not resolve whether this reflects a different threshold or a loose description of the same flag. Conservation statements from these projects should therefore not be pooled without checking the definition. _3,1

Possible Reconciliations

• Hypothesis 1 (loose wording): The structural-annotation project may have read the same is_core column as the other projects and paraphrased it imprecisely as “majority”. If so, its results share Side A’s boundary and the disagreement is only one of description.
• Hypothesis 2 (different operational definition): That project may have derived or recomputed a core label with a majority cutoff instead of using the stored flag. If so, its core set could differ from one defined by Side A’s threshold, and its core-versus-accessory contrasts would not be directly comparable to those using that threshold.
• Hypothesis 3 (mixed usage): Some steps may have used the stored flag and others a majority-based summary. If so, each result would need to be checked separately to establish which definition it rests on.

Resolving Work

• Data: the structural-annotation project’s notebooks and queries. Method: trace whether core labels come directly from the stored is_core column or are recomputed from genome counts. Question: which definition do its core-versus-accessory results actually use?
• Data: the pangenome gene_cluster table. Method: for each species, compute observed prevalence of clusters flagged is_core and compare it with the ≥95% threshold and with a majority cutoff. Question: does the stored flag behave as Side A states?
• Data: the structural-annotation project’s core-versus-accessory comparisons. Method: rerun them under both a ≥95% and a majority definition. Question: do its conclusions change with the definition, so that pooling with Side A projects is unsafe?
• Data: species clades with few genomes in the pangenome tables. Method: tabulate how many clusters change classification between the two definitions as clade size varies. Question: is any definitional mismatch amplified where clade-size bias is already a concern?