Phage characterization analysis in respiratory samples from infected patients based on metagenomic next-generation sequencing

Background Respiratory tract infections are common infectious diseases, with microbial dysbiosis closely linked to clinical outcomes in the host. As key regulators of bacteria, phages can influence the structure and stability of microbial communities by infecting host bacteria. Metagenomic next-generation sequencing (mNGS) enables comprehensive analysis of phage community characteristics in clinical samples. Methods This study included 6,404 clinical samples, comprising 4,837 bronchoalveolar lavage fluids (BALF) and 1,567 sputum samples, for metagenomic next-generation sequencing (mNGS), while collecting patient demographics, sample types, mNGS results, and clinical outcomes. Host-derived sequences were removed post-sequencing and aligned against viral reference databases. Phage community structures across sample types were assessed using alpha and beta diversity metrics. Spearman correlation analysis explored associations between phages and bacteria. Further bioinformatics analysis was performed on 194 samples, including viral sequence assembly and identification using SPAdes, VirSorter2, and PhaMer; CD-HIT clustering and redundancy removal; CheckV quality assessment; PhaTYP lifestyle prediction; Prodigal protein gene annotation; and BLASTP alignment against the CARD database to screen for phage resistance genes. Results The sputum and BALF groups exhibited comparable richness, diversity, and evenness, yet their community structures differed significantly. Intensive Care Unit (ICU) admission status was closely associated with reduced phage community diversity and significant alterations in community structure, and the abundance distribution of several phage families ( Peduoviridae , Autoscriptoviridae , Casjensviridae , Demerecviridae ) also changed significantly. Additionally, the phage community structure in sputum samples was significantly associated with patient clinical outcomes. Correlation analysis demonstrated that the Aliceevansviridae family in sputum samples had extensive positive associations with various bacteria. After assembly, 69.5% of pOTUs were predicted to be temperate phages, and 28.9% were predicted to be virulent phages; moreover, the vast majority (99.2%) of phage sequences showed low similarity to antibiotic resistance genes. Conclusion This study identifies distinct phage community characteristics across respiratory sample types and reveals that ICU patients exhibit reduced phage diversity and markedly altered community structures. Furthermore, the phage composition in upper respiratory tract samples shows a clear relationship with patient prognosis, providing new insights into respiratory infection microecology.