Rapid evolution is a defining feature of many viruses. This evolution is often detrimental to human health, as it enables viruses to escape from immunity and drugs. However, it also offers a unique vantage point to study basic questions in evolutionary biology, since we can watch viruses change in almost real time. Our lab studies virus evolution to address a range of basic and medical questions. Below are brief descriptions of our current major topics of interest, along with links to key papers.
It is now quite easy to monitor virus evolution at the genetic level using sequencing. But what do all of those mutations mean in terms of viral fitness and antigenicity? To address this question, we have developed ways to measure the effects on viral growth of all amino-acid mutations to key proteins from viruses such as influenza and HIV. We have also extended these techniques to map how all mutations affect viral neutralizaton by antibodies. These methods can quantify the potential for a virus to escape different antibodies, inform the design of vaccines, and help with understanding implications of viral mutations observed in infected humans. We are also examining whether these methods can be leveraged to better forecast viral evolution in nature.
New experiments like the ones described above provide a wealth of detailed information about how mutations affect specific genes. How can we use this information to better model evolution in nature? We have developed a computational framework to achieve this goal by using experiments to inform more accurate phylogenetic substitutions models. We have shown that this approach can better identify sites of positive selection during protein evolution, and have created software that makes the methods easy to use.
Even nominally "wildtype" stocks of virus are in fact extremely genetically diverse, and it is easy to imagine how this viral diversity might affect the outcome of infection. To examine the extent that this is the case, we initially applied single-cell mRNA-seq techniques to characterize the diversity of transcriptional outcomes in single influenza-infected cells. This initial work revealed that the outcome of infection indeed varied widely across cells, and that infected cells only rarely triggered innate-immune pathways. However, standard single-cell mRNA-seq is insufficient to determine if viral mutations are responsible for the heterogeneity in infection outcome, since it does not determine the full sequence of the infecting virions for each cell. We have therefore developed a new approach that leverages long-read PacBio sequencing to fully sequence the virions that infect single cells, and have shown that viral mutations indeed account for a substantial amount of the heterogeneity in infection outcome and innate-immune induction.
Viruses such as influenza undergo selection at multiple different scales. They replicate and generate mutations during infections of single hosts, but in order for a mutation to spread it then needs to be successfully transmitted to new hosts. We have used deep sequencing of long-term human infections to assess the concordance of selection at these two very different scales. We are also actively involved in the computational re-analysis of existing deep sequencing data to resolve key questions about influenza virus evolution.
Our most long-standing interest has been understanding how molecular properties are related to more abstract evolutionary properties such as evolvability and epistasis. We have shown that in some cases the epistasis that is observed on multi-step evolutionary trajectories of viruses can be rationalized in terms of measurable properties such as protein stability. We are also using experiments that study mutations in different genetic backgrounds to help build a more accurate picture of the extent and basis of epistasis.
Much of our research involves computational analyses of relatively large datasets. We are therefore interested in reproducible research. In addition to posting our manuscripts on bioRxiv and putting our computer code on GitHub, we've been trying other approaches to make it easy for others to reproduce and extend our work. We implement our common computational analyses in documented software packages, and post detailed descriptions and lab notes for experimental methods (for instance, here, here, and here). We have also been endeavoring to create repositories that include all relevant text, code, and data for individual papers---spanning from the initial writing to the final published version. Here are some examples from our recent papers on single-cell virus sequencing, mapping escape mutations for anti-influenza antibodies, and assessing how experimentally informed site-specific models affect phylogenetic estimates of viral divergence times. Our hope is that making available everything that went into these papers will allow others to better understand, assess, and extend our work.