Quantifying Pathogen Surveillance Using Temporal Genomic Data

Chan, Joseph M.; Rabadan, Raul

Currently, there is no standard quantitative measurement of the appropriate level of genomic surveillance of a given pathogen. One immediate approach might be to consider the absolute number of sequenced isolates, but this method fails to account for the diversity and time information of the population sampled. A comparatively large number of sequences may be insufficient to capture high genetic diversity in a pathogen population. Another possible strategy appeals to clustering techniques. After representing sequences as points in a mutational landscape, a highly clustered structure could potentially reflect highly biased sampling. Once again, this method ignores time and can be confounded by evolutionary processes like bottlenecks. A final tactic might be to mea-sure the genetic diversity within a sample of the pathogen. Phylogenetic reconstruction and tallying of species richness is one method that characterized the subclades of avian H5N1 (8) , but such analyses are cumbersome, with arbitrary boundaries for classifying clades. Techniques more grounded in information theory include Shannon's entropy and Rao's quadratic entropy (22) , but high diversity in a sample does not necessarily correlate with high surveillance either.