Investigate effects of forecast horizons and submission lags at antigenic positions

[huddlej]

Investigate how changes in forecast horizons and submission lags change the distance between predicted and observed future populations at antigenic positions (i.e., "epitope" positions from Wolf et al. 2006) to understand whether we see the same effects with that subset of positions as we see across all HA positions.

• Calculate the earth mover's distance between natural H3N2 populations using only amino acid sequences from epitope positions and focusing on forecasts from the empirical retrospective fitnesses.
• Summarize the effects of realistic interventions using these distances between epitope sequences, describing how improvements in forecasting accuracy differ when we only consider sites under antigenic evolution.