Transmission network inference from molecular data is inherently imprecise, yet certain features of inferred networks are both robust to inference errors and informative. We will present a high throughput framework for inferring molecular transmission networks from HIV-1 surveillance and drug resistance testing data (HIV-TRACE). Using published analyses of inferred molecular national (US), and local (New York City, San Diego) transmission networks, we will discuss several statistical approaches to answering the following questions
1). How useful is the binary attribute (clustered/non-clustered) for characterizing the dynamics of HIV-1 transmission, and rates of transmission between risk groups?
2). Which molecular network features are useful for targeting surveillance and prevention efforts?
3). How can one improve the reliability of inferring individual edges (transmission links)?
4). Can simulation based inference strategies (e.g. agent-based or ABC) be coupled with imprecise network inference to estimate epidemiologically relevant parameters?