HMM likelihood for haplotype reconstruction

Prabhav suggested using a HMM to describe the likelihood of an observed viral sequence. This allows for dependence between the states, and also for, i.e., increased probability of error after repeats.

Finding different epochs in a DPM

A DPM is like a collection of poisson processes. We have data drawn from these processes which also contains covariates, such as case/control, age, date.

Use a random forest to fit the counting processes.