Quantifying Pathogen Surveillance Using Temporal Genomic Data

Chan, Joseph M.; Rabadan, Raul

Given a viral evolutionary rate of substitutions per site per year, R ϭ t ϫ is thus the expected proportion of mutations that have accumulated over the span of t years. In this report, we examine short time periods on the order of a decade and do not expect the true genetic distance to diverge greatly from the Hamming distance. However, we calculate the q2 coefficient by using a number of distance methods, including Hamming distance, Jukes-Cantor, Kimura, Tamura, Tajima-Nei, Hasegawa, and Nei-Tamura, and find little significant change (see Results). If the surveillance of a pathogen in a given time and place is sufficient, we would therefore expect a maximal number of viral isolates to have a closest ancestor in the sequence database with a sequence identity of greater than 1 Ϫ R. For the q coefficient, we chose to use t ϭ 2 years since antigenic variant strains of influenza virus emerge and become predominant over a period of roughly 2 to 5 years (23); however, any value of t can be used. One consideration is that substitution rates can vary over time and across different lineages. To incorporate such variability into our calculation of the q2 coefficient, we used the 95% highest posterior density (HPD) interval of the evolutionary rate collected from the literature (see Table S1 in the supplemental material) (24) (25) (26) .