ABSTRACT Adenoviral keratoconjunctivitis is one of the most common conditions in all of medicine. Despite being a common cause of morbidity world-wide, there are no known host or pathogen factors that predict clinical outcomes in this condition. A recent, large, international clinical study of adenovirus-related conjunctivitis revealed that approximately 10% of patients suffer from long term visual loss. A limited deep DNA sequencing study of AdV D8 clinical samples conducted under our previous R21 (1R21EY027453) revealed unexpected sequence diversity in the viral genome, with approximately 600 sequence variants among 87 samples within the same hexon-defined molecular type. These variants assorted into three subtypes with different propensity to poor outcome. Remarkably, using machine learning approaches, we found we were able to predict one critical outcome – the development of subepithelial infiltrates – from knowledge of the viral sequence alone. Our previous study sequenced samples from the placebo arm of the NVC-422 clinical trial. We have an additional 157 AdV D8 samples that have not been sequenced. In Aim 1 we propose sequencing the adenovirus of these samples in order to a.) further characterize the sequence diversity of AdV D8 and b.) validate our machine learning method for predicting development of subeptithelial infiltrates While AdV D8 was the most prevalent cause of AKC in our study worldwide, unexpectedly we found in the United States that AdV E4 was the most prevalent type. In Aim 2, we propose fully sequencing all 36 samples in the study from type E4, as well as 23 samples from type B3, 9 samples from D19, and a total of 35 samples distributed between type D53, D56, and D64 to determine their molecular diversity, and to apply the same machine learning methods to this set of samples to determine if outcomes can be predicted from viral sequence variants for these types as for AdVD8. The results of these studies will expand our understanding of the molecular pathogenesis of viral conjunctivitis, and will provide biomarkers for predicting outcomes from this condition. These advances will facilitate future efforts toward developing therapies for this common condition.