Estimating confidence intervals for Freyja outputs

[ArtPoon]

Freyja gives users the option of running bootstrap samples to assess confidence in frequency estimates. However, running this with 100 bootstraps is taking too long. We should investigate running a smaller number (say 10 bootstraps) and then making a rough estimate of the confidence interval based on the standard error for frequency estimates.

[ArtPoon]

I had proposed a crude estimate based on a small number of bootstrap replicates, and applying a logit transform to a normal approximation of 95% CI - will transfer details from Slack thread to here

[ArtPoon]

look into incorporating virus density estimates from hub database and using nonparametric sampling from hypergeometric distribution to estimate confidence intervals