Contributions
Pan-Bacterial Anti-Phage Defense Arsenal analyzed seven anti-phage defense-system families across the KBase Data Lakehouse pangenome 1. It detected markers in 27,626 of 27,690 species-level pangenomes (99.8%), yielding 930,573 marker hits across 825,476 unique gene clusters 1. Its arms-race and syndrome tests used a quality-filtered set of 7,323 species with ≥5 genomes 1.
- Prevalence. Computed across all 27,626 pangenome species with at least one defense hit, species-level prevalence ranged from 7.2% for CBASS to 96.1% for CRISPR-Cas 1. EggNOG description-based CRISPR detection gave 96% prevalence, whereas the specific Cas1 Pfam marker PF01867 gave approximately 55% on the same pangenome, so the project treated the 96% value as an upper bound 1.
- Defense count and prophage burden. Defense-system count correlated with prophage-cluster burden at Spearman ρ = 0.609 (n = 7,323) 1. After residualizing on log10 median genome size and phylum, the partial ρ was 0.301 (p = 1.6 × 10⁻¹⁵³) 1. All nine major phyla showed positive, significant partial correlations, ranging from ρ = 0.530 in Campylobacterota to ρ = 0.185 in Actinomycetota 1. The report’s narrative gives Actinomycetota as ρ = 0.18 (p < 10⁻⁷), which conflicts with its results table (p = 1.3×10⁻⁷) and is left unreconciled 1. The project describes this as an association with prophage-cluster burden, not a direct measurement of phage pressure 1.
- Syndromes. Using phylum-stratified permutation nulls and Benjamini–Hochberg false-discovery-rate (BH-FDR) correction, 27 of 28 system pairs showed significant positive co-occurrence at q < 0.05 1. The strongest pair was restriction-modification (R-M) Type II × Gabija (z = 46.1, odds ratio (OR) = 24.0) 1. CRISPR-Cas × CBASS was the only non-significant pair (z = 0.21, p_emp = 0.98) 1. The mechanistic rationale for the R-M Type II × Gabija pairing was cited from prior work and not tested 1.
- Accessory-genome enrichment. Six of seven systems were enriched in the auxiliary and singleton pangenome relative to a background of 46.8% core and 37.9% singleton 1. The project noted this is consistent with, but not a direct demonstration of, horizontal transfer 1. DISARM was near baseline and was treated as unreliable, likely a detection artefact of the widespread SNF2 helicase Pfam anchor PF00176 1.
Prophage-AMR Co-mobilization Atlas examined associations between antimicrobial-resistance (AMR) genes and prophage markers across the GTDB pangenome 2. It identified 83,008 AMR gene clusters and 3,465,244 broad prophage marker clusters; 14,669 of 27,702 species (52.9%) carried both 2.
- Species level. Across 4,770 species, prophage marker density correlated with AMR repertoire breadth at Spearman rho=0.572 (p<10^-300) 2. The partial correlation controlling for genome count was rho=0.464 (p=1.0×10^-253), and the association was significant in all five reported major phyla 2.
- Gene level. In the top-100 AMR-burdened species, sampled at 20 genomes per species (1,953 genomes total) rather than all available genomes, AMR genes within 10 genes of a prophage marker were 67.6% accessory versus 65.5% for distal genes (odds ratio 1.10, one-sided p=0.005) 2. This effect was modest and threshold-dependent: the odds ratio was 0.78 at 3 genes and 1.28 at 50 genes 2. Only 33 of 74 testable species had odds ratios above 1 2.
- Scale dependence. The project found the prophage–AMR association strong at species level but modest, threshold-dependent and heterogeneous at gene-neighborhood scale, a contrast that feeds scale-dependent-mobile-element-associations 2.
- Fitness-cost null. The planned fitness-cost comparison could not be run because RB-TnSeq (random barcode transposon sequencing) data cover only 48 model organisms with poor overlap with the analyzed species 2.
- Caveats. The project stated that its results are correlative and do not establish phage-mediated AMR transfer 2. Gene distances were ordinal rather than base-pair distances 2. Prophage markers came from keyword and Pfam matching rather than dedicated prophage predictors such as PHASTER or geNomad, so they may include phage-defense false positives and may miss divergent prophages 2.
Ecotype Functional Differentiation clustered auxiliary gene presence/absence matrices with PCA (principal component analysis) followed by KMeans clustering, searching k = 2–6, to test whether within-species ecotypes differ in COG (Clusters of Orthologous Groups) profiles 3. It found valid ecotypes in 12 of 15 sampled species (80%), covering 1,820 genomes, with a mean silhouette score of 0.215 that indicates overlapping rather than discrete boundaries 3.
- Overall differentiation. Of 257 tests, 170 (66.1%) were significant after BH-FDR correction at q < 0.05, and all 12 species had at least one differentiated COG category 3.
- Adaptive versus housekeeping categories. The hypothesis that adaptive categories differentiate more than housekeeping categories received partial support 3. Significance rates were 79.8% versus 68.8%, a 1.16x ratio 3. Mean effect sizes were 0.0136 versus 0.0064, a 2.13x ratio (one-sided Mann–Whitney U p = 2.53 x 10^-6) 3.
- Largest effects. Category S (unknown function) had the largest mean effect size, 0.0392, and category L (replication, recombination and repair) had the second largest, 0.0337 3. The project described these effects as small, approximately 3–4 percentage-point differences in COG-category proportions, and warned that statistical significance may partly reflect large sample sizes 3.
- Caveats. The project reported several limitations 3:
- The report states that the assigned genomes span 6 phyla but names only four 3.
- About 62% of gene clusters lacked COG annotations, so unannotated ecotype-specific adaptive genes may be missed 3.
- KMeans was used because HDBSCAN (a density-based clustering method) was unavailable on the cluster; KMeans assumes spherical clusters and requires a selected k, whereas HDBSCAN could better handle variable-density subpopulations 3.
- Without within-species phylogenetic controls, ecological adaptation cannot be distinguished from phylogenetic or demographic substructure 3.