AbstractStudying in vivo fitness costs of mutations in viruses provides important insights into their evolutionary dynamics, which can help decipher how they adapt to host immune systems and develop drug resistance. However, studying fitness costs in natural populations is difficult, and is often conducted in vitro where evolutionary dynamics differ from in vivo. We aimed to understand in vivo fitness costs of mutations in Hepatitis C virus using next generation sequencing data. Hepatitis C virus is a positive-sense single-stranded RNA virus, and like many RNA viruses, has extremely high mutation and replication rates, making it ideal for studying mutational fitness costs. Using the ‘frequency-based approach’, we estimated genome-wide in vivo mutation frequencies at mutation-selection equilibrium, and inferred fitness costs (selection coefficients) at every genomic position using data from 195 patients. We applied a beta regression model to estimate the effects and the magnitudes of different factors on fitness costs. We generated a high-resolution genome-wide map of fitness costs in Hepatitis C virus for the first time. Our results revealed that costs of nonsynonymous mutations are three times higher than those of synonymous mutations, and mutations at nucleotides A/T have higher costs than those at C/G. Genome location had a modest effect, which is a clear contrast from previously reported in vitro findings, and highlights host immune selection. We inferred the strongest negative selection on the Core and NS5B proteins. We also found widespread natural prevalence of known drug resistance-associated variants in treatment naive patients, despite high fitness costs of these resistance sites. Our results indicate that in vivo evolutionary patterns and associated mutational costs are dynamic and can be virus specific, reinforcing the utility of constructing in vivo fitness cost maps of viral genomes.Author SummaryUnderstanding how viruses evolve within patients is important for combatting viral diseases, yet studying viruses within patients is difficult. Laboratory experiments are often used to understand the evolution of viruses, in place of assessing the evolution in natural populations (patients), but the dynamics will be different. In this study, we aimed to understand the within-patient evolution of Hepatitis C virus, which is an RNA virus that replicates and mutates extremely quickly, by taking advantage of high-throughput next generation sequencing. Here, we describe the evolutionary patterns of Hepatitis C virus from 195 patients: We analyzed mutation frequencies and estimated how costly each mutation was. We also assessed what factors made a mutation more costly, including the costs associated with drug resistance mutations. We were able to create a genome-wide fitness map of within-patient mutations in Hepatitis C virus which proves that, with technological advances, we can deepen our understanding of within-patient viral evolution, which can contribute to develop better treatments and vaccines.