Effective control of bacterial infections remains a significant challenge in aquaculture. The marine bacterium Piscirickettsia salmonis (P. salmonis), responsible for piscirickettsiosis, causes widespread infections in various salmon species, leading to substantial mortality and economic losses. Despite efforts to genetically characterize P. salmonis, critical gaps persist in understanding its virulence factors, antimicrobial resistance genes, and single-nucleotide polymorphisms (SNPs). This study addresses these gaps through a comparative analysis of the pan-genome and core genomes of 80 P. salmonis strains from different geographical regions and genogroups. P. salmonis had an open pan-genome consisting of 14,564 genes, with a core genome of 1257 conserved genes. Eleven virulence-related genes were identified in the pan-genome, categorized into five functional groups, providing new insights into the pathogenicity of P. salmonis. Unique SNPs were detected in four key genes (gyrA, dnaK, rpoB, and ftsZ), serving as robust molecular markers for distinguishing the LF and EM genogroups. Notably, AMR genes identified in four LF strains suggest evolutionary adaptations under selective pressure. Functional annotation of the core genomes using the gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases demonstrated conserved gene clusters linked to essential intracellular survival mechanisms and bacterial pathogenicity. These findings suggest a direct association between core genome features and variations in pathogenesis and host–pathogen interactions across genogroups. Phylogenetic reconstruction further highlighted the influence of AMR genes on strain divergence. Collectively, this study enhances the genomic understanding of P. salmonis and lays the groundwork for improved diagnostic tools and targeted therapeutics to manage piscirickettsiosis in aquaculture.

Fish nocardiosis, caused by Nocardia seriolae , poses a persistent threat to the aquaculture industry. Yet, the genomic determinants underlying strain-level diversity and adaptation remain poorly understood due to the high nucleotide conservation of this species. The objective of this study was to characterize genome-level variation among all publicly available complete genomes using integrative comparative genomics approaches that extend beyond nucleotide identity metrics. Nine complete genomes were analysed using average nucleotide identity, whole-genome structural rearrangement analysis, single-copy phylogenomics, genomic island prediction, pangenome reconstruction, functional annotation and antimicrobial resistance and virulence profiling. Although average nucleotide identity values confirmed extreme nucleotide conservation across all strains, extensive strain-specific structural rearrangements, including inversions, translocations and duplications, were detected. Phylogenomic reconstruction resolved geographically associated lineages despite minimal nucleotide divergence. Pangenome analyses supported an open pangenome dominated by a large, conserved core genome with limited but persistent accessory gene content and core biased gene duplication. Functional profiling revealed enrichment of transcriptional and metabolic processes, while resistance and virulence analyses identified only conserved intrinsic determinants shared across all strains. These findings demonstrate that genome architecture and pangenome dynamics provide critical resolution for understanding N. seriolae diversification. The study highlights the importance of integrating structural genomics and phylogenomics for strain tracking and surveillance in aquaculture systems.

Smishing is a social engineering attack using SMS containing malicious content to deceive individuals into disclosing sensitive information or transferring money to cybercriminals. Smishing attacks have surged by 328%, posing a major threat to mobile users, with losses exceeding \$54.2 million in 2019. Despite its growing prevalence, the issue remains significantly under-addressed. This paper presents a novel hybrid machine learning model for detecting Bangla smishing texts, combining Bidirectional Encoder Representations from Transformers (BERT) with Convolutional Neural Networks (CNNs) for enhanced character-level analysis. Our model addresses multi-class classification by distinguishing between Normal, Promotional, and Smishing SMS. Unlike traditional binary classification methods, our approach integrates BERT's contextual embeddings with CNN's character-level features, improving detection accuracy. Enhanced by an attention mechanism, the model effectively prioritizes crucial text segments. Our model achieves 98.47% accuracy, outperforming traditional classifiers, with high precision and recall in Smishing detection, and strong performance across all categories.

Atlantic salmon (Salmo salar) aquaculture suffers substantial losses due to infectious and non-infectious diseases. However, the ecological basis of gut microbiome dysbiosis remains poorly understood across geographically distinct production systems. A key unresolved question is whether dysbiosis results from the restructuring of resident microbial communities or from the introduction or acquisition of exogenous microorganisms. To address this, a multi-country pooled cross-study reanalysis of publicly available 16S rRNA gut microbiome datasets from Ireland (V1V2, n = 234), Norway (V1V3, n = 4), Chile (V4, n = 70), and China (V4, n = 12) was conducted, comprising 320 samples after quality check and taxonomic filtering. An integrative analytical framework was applied, combining compositional microbiome analysis, microbial source tracking, and machine learning approaches to identify generalized disease-associated dysbiosis signatures in a pooled dataset comprising geographically distinct salmon populations. The analysis found strong between-cohort structure in microbiome composition, whereas disease status produced smaller but significant compositional differences without clear visual separation. Alpha diversity was not significantly linked to disease condition, although ANCOM-BC2 analysis identified 48 genera with significantly altered abundances between healthy and diseased fish. In the pooled dataset, healthy fish showed higher abundance of genera such as Mycoplasma, Oleispira, Aliivibrio, Psychrobium, and Cobetia. In contrast, diseased individuals were enriched in Pseudomonas, Prevotella, Massilia, Chryseobacterium, and multiple genera affiliated with Flavobacteriaceae. Using Random Forest classification, an AUC-ROC of 0.819 was achieved, with Oleispira the strongest health-associated genus in both the Shapley Additive Explanations (SHAP) ranking and the ANCOM-BC2 analysis; this association derives from the Irish cohort, whereas Mycoplasma, Pseudomonas and Aliivibrio retained direction and significance when the Irish cohort was excluded. Under leave-one-study-out validation, accuracy was 66.2% for Ireland, 58.3% for China, 50.0% for Norway and 35.7% for Chile. Classifier performance, therefore, showed limited cross-cohort transferability, and the pooled estimate should not be interpreted as a diagnostic accuracy estimate. Microbial source tracking indicated substantial compositional overlap between diseased fish microbiomes and the selected healthy reference pool, with a mean healthy source attribution of 93.5%; this does not exclude acquisition of organisms from external sources. Overall, these findings suggest that gut dysbiosis in Atlantic salmon is consistent with compositional reconfiguration of a shared genus pool, although this pattern does not establish microbial provenance.