One of the most important unknowns in the #nCoV2019 epidemic is the total number of _infections_ as opposed to _cases_, as there may be many mild infections that do not rise to case definition. I sought to address this with phylodynamic methods here: bedford.io/projects/ncov-… 1/9
Here, I started by following @arambaut's work (virological.org/t/phylodynamic…) and set up beast.community to estimate viral population dynamics from sequence data alone. 2/9
@arambaut This measures how quickly observed genomes share ancestry to estimate the rate of exponential growth, arriving at an estimated doubling time of 7.2 days (95% CI 5.0-12.9), inline with previous modeling estimates from case data. 3/9 
@arambaut I convert estimates of coalescence to estimates of pathogen prevalence following phylodynamic methods described here en.wikipedia.org/wiki/Viral_phy…, assuming a range of heterogeneity in secondary transmissions ala biorxiv.org/content/10.110… 4/9
@arambaut This estimates a median prevalence on 8 Feb of 28,500 currently infected with a 95% uncertainty interval of between 7500 and 104,300 currently infected. 5/9
@arambaut And further estimates a median total incidence on 8 Feb of 55,800 total infections since start of epidemic with a 95% uncertainty interval of between 17,500 and 194,400 total infections. 6/9
@arambaut This approach estimates an infection-to-case reporting rate of between 18% and 100%. Although there are obviously wide uncertainty intervals, I believe it is safe to conclude that case reporting is largely in line with expectations given severity. 7/9
@arambaut The wide range of 17,500 to 194,400 total infections comes from a combination of uncertainty in effective population size (limited by amount of current sequence data) and uncertainty in degree of heterogeneity in secondary transmissions. 8/9
@arambaut This analysis has relied on open sharing of nCoV pathogen genomic data by research groups from all over the world. I gratefully acknowledge their contributions. I'll try to update this as more data is generated and shared. 9/9
