Large-Scale Comparative Genomics of 1,374 Lactiplantibacillus plantarum Genomes Reveals an Open Pan-genome and Strain-Level Variation in Probiotic-Associated and Safety-Related Genomic Features.

Lactiplantibacillus plantarum is widely studied for food fermentation and probiotic applications, but its functional and safety-related characteristics are strongly strain dependent. Large-scale comparative genomics can characterize this diversity and support genome-informed strain prioritization, although genomic evidence alone does not establish probiotic efficacy or safety. A RefSeq genome collection of L. plantarum was subjected to accession-level auditing, exact-sequence dereplication, and predefined genome-quality filtering, yielding 1,374 unique genomes for final analysis. Genome characteristics, pan-genome structure, probiotic-associated annotation proxies, targeted sequence evidence, predefined safety-related genomic markers, isolation-source metadata, and COG functional annotations were evaluated. The final cohort had a mean genome size of 3.286 ± 0.126 Mb and a mean GC content of 44.419 ± 0.168%. Genome size and GC content showed a negative dataset-level association (Pearson's r = - 0.767; R2 = 0.588). The observed pan-genome comprised 32,500 gene families, including 1,175 core and 31,325 accessory families, and remained non-saturating, with a Heaps' law exponent of γ = 0.289. Probiotic-associated annotation proxies showed heterogeneous prevalence, and targeted sequence analysis demonstrated that annotation nomenclature could substantially influence apparent marker detection. Evidence-supported antibiotic-resistance markers were detected in 46 genomes (3.35%) and mobile-genetic-element-associated markers in 93 (6.77%), whereas ldhD was detected in 1,372 genomes (99.85%). The predefined descriptive concern score assigned 1,236 genomes (90.0%) to the Lowest Genomic Concern category, 136 (9.9%) to the Low category, and two (0.1%) to the Moderate category. Only 1,327 of the 32,500 gene families received a recognized primary COG assignment. The quality-filtered L. plantarum RefSeq collection exhibits an open, accessory-rich pan-genome and substantial strain-level genomic diversity. Annotation-dependent marker detection, uneven database representation, and the limited discrimination of the descriptive concern score emphasize that genome-based screening should be used for candidate prioritization rather than as evidence of probiotic efficacy or definitive safety. Strain-specific phenotypic and application-relevant safety validation remains essential.