CheckM estimates genome completeness and contamination. The two projects here use it in three ways. They filter genomes by quality before comparing cohorts. They rescale genome-content measures by completeness. They also compare mean completeness between cohorts to test whether differences in genome content are artefacts of MAG (metagenome-assembled genome) quality 1,2.
Use in Bacillota B subsurface genome comparisons
The project compared deep-clay bacillota-b anchor genomes (n=10) with soil-baseline Bacillota_B genomes (n=62). After CheckM rescaling, mean genome size in base pairs was 4,323,230 for the anchors and 3,233,715 for the baseline. The effect size was Cohen’s d +1.37 (Cohen’s d is a standardized effect size; see cohens d), with Mann–Whitney p 0.013. Mean CheckM-rescaled eggnog OG (orthologous group) counts were 2,771 for the anchors and 2,233 for the baseline (d +1.32; p 0.009). In both measures the deep-clay anchors were larger 1.
Mean CheckM completeness was nearly the same in the two cohorts: 94.7% for the anchors and 94.3% for the baseline (d +0.08; p 0.93). The project treats this null result as a critical control and concludes that the size difference cannot be a MAG-quality artefact. On the evidence presented, the matched completeness weakens a quality artefact as an explanation for the genome-size difference 1.
The project’s H2 analysis table has 6 rows. It applies Wilcoxon tests and Cohen’s d to genome size, OG count, GC, and CheckM measures. These findings feed subsurface-bacillota-specialization 1.
Use in clay-confined subsurface cohorts
The clay-confined subsurface project built its cohorts from kbase-ke-pangenome NCBI environment metadata. It selected genomes whose isolation_source or env_* fields contained clay-related keywords, then joined them to the pangenome by biosample ID. The final cohorts were filtered at CheckM ≥80% completeness and ≤5% contamination 2.
Quality filtering shrank the deep-anchor cohort from 9 genomes to 6. Before filtering, this cohort held 8 mont-terri Opalinus borehole genomes (BRC-3 + BIC-A1) and 1 bentonite-formation genome. Because so few genomes remain after filtering, results for this cohort rest on a very small sample 2.
The project’s self-sufficiency test compared mean counts of complete GapMind amino-acid pathways between cohorts. After CheckM ≥80 filtering, the 6 deep anchors had a lower mean count than the 137 baseline genomes: 15.50 versus 17.14 (d=−0.84, p=0.009). The 30 shallow anchors had a higher mean count than the same 137 baseline genomes: 17.87 versus 17.14 (d=+0.43, p=0.029). These comparisons bear on biosynthetic-self-sufficiency-and-cultivation 2.
The project gives two caveats for these pathway counts. First, the 18-pathway GapMind universe creates a ceiling effect: most cultured organisms, deep or shallow, reach 17–18/18 regardless of habitat, which leaves little room for a positive signal at the upper end. Second, the cultured cohort lacks the extreme self-sufficient lineages described in the literature, which are typically uncultivated and recovered as MAGs or single-cell genomes 2.
The figure h1_self_sufficiency_violin.png shows how the counts of complete amino-acid pathways are distributed in each cohort, both before and after CheckM filtering 2.
Caveats
CheckM filtering and rescaling correct for differences in genome completeness. On their own, they do not show that the remaining contrasts in genome content are biological 1,2.
Only the Bacillota B project reports matched mean CheckM completeness between the cohorts it compares, and uses that match as evidence against a quality artefact. The clay-confined project reports quality filtering, not a comparison of completeness between cohorts 1,2.