Recurrent intrahost G-gene haplotype structures suggest potential co-transmission in RSV-A

Respiratory syncytial virus (RSV) is a leading cause of infant hospitalization worldwide. Its molecular epidemiology rests on consensus G-gene sequences, which capture only the dominant virus in each infection. Deep sequencing reveals far more variation inside each host. Linking those variants to one another gives a finer view of transmission. Here we reconstructed G-gene haplotypes from two independent RSV-A deep-sequencing datasets (Australia, n = 37; United States, n = 72). We phased variants using read-level linkage, then placed each haplotype on a global phylogeny. More than 90% of samples carry multiple haplotypes, which often sit on distinct phylogenetic branches. One cross-branch pair dominates each dataset: C4+C6 in 23 of 37 Australian samples (62%), and C18+C21 in 26 of 72 US samples (36%). Both pairs recur far more often than month- and region-matched permutations predict (both P = 5.0 × 10^-5). Mutation-probability calculations suggest that recurrent within-host de novo mutation is unlikely to fully account for these patterns, particularly for the five linked sites in C4+C6 (probability ≤ 5.36 × 10^-15). Repeated joint introduction of pre-existing haplotypes offers a simpler explanation, and transmission-chain sampling can now test it directly. All branch-defining sites mapped to G-HVR2. In the examined global consensus G-gene set, dominant-branch states appeared, while minor-branch combinations were not detected under our criteria. Haplotype-resolved analysis thus turns sub-consensus diversity into linked, interpretable epidemiological signal.