Inference and Theory of Viral Fitness Landscapes to Predict Influenza A and HIV Evolution

• Applicant: Dr. Nikhil Sharma, Ph.D.
• Subject Area: Bioinformatics and Theoretical Biology ; Epidemiology and Medical Biometry/Statistics ; Evolution, Anthropology
• Term: since 2026
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 579803321

According to the WHO, influenza viruses cause between 290,000 and 650,000 deaths globally each year. Among children under five, 99 percent of influenza-related respiratory deaths occur in developing countries. HIV is another major global health issue, having caused around 44.1 million deaths to date. By the end of 2024, an estimated 40.8 million people were living with HIV, with about 65 percent residing in African countries. In 2024 alone, approximately 630,000 people died from HIV-related causes. These viruses also place a heavy economic burden on society. In the U.S.A., the annual cost of influenza A is estimated at 11.2 billion US dollars, covering both direct medical and indirect costs. For HIV, the average lifetime medical cost per person in the U.S.A. is estimated at around 420,000 US dollars. To help address these challenges, we propose a project that aims to forecast the evolution of Influenza A and HIV. Our goal is to build a predictive pipeline that takes, as input, data on circulating viral strains and their response to the human immune system. The output will be predictions about which strains are likely to escape immune response and spread more efficiently. While the focus is on Influenza A and HIV, the same pipeline can be adapted for other viruses like SARS-CoV-2. As both viruses and immune systems evolve, the pipeline can be reused with updated experimental data. This interdisciplinary project lies at the intersection of molecular evolution, theoretical population genetics, and epidemiology. It involves three key steps. First, we will use molecular data to construct fitness landscapes for the viruses. Second, we will apply evolutionary theory to estimate how quickly viruses adapt. Third, we will simulate viral genealogies on these landscapes to predict short-term evolutionary trends. We will use existing deep mutational scanning (DMS) data, in which several virus strains are competed in the lab against cocktails of antibodies. The performance of each strain is recorded as a set of values in the DMS data. From this, we will infer the fitness landscapes of Influenza A and HIV. These landscapes act as roadmaps for viral evolution, visualized as movement across fitness peaks, valleys, and ridges. Using methods from population genetics, we will then estimate how long it takes for key mutations - those enabling immune escape and higher transmissibility - to appear. Finally, we will simulate viral genealogies on these landscapes to forecast how these viruses may evolve in the near future. Understanding which virus strains are most likely to become dominant helps guide timely public health responses. For influenza, this means updating seasonal vaccines to better match the circulating strains. Similarly, by improving our understanding of HIV evolution, the proposed project supports the long-term goal of developing an effective HIV vaccine.

• DFG Programme: Fellowship
• International Connection: USA
• Host: Professor Joshua Plotkin, Ph.D.