[Mandelbrot2004 (OP)]

People who recovered from COVID in Korea have been reinfected (see Korea Herald or South China Morning Post). I've heard anecdotally that the virus is mutating fast, e.g. it was different in Southern Italy by the time it was transmitted from Northern Italy.

What evidence do we have for speed of mutation (an article in healthline dot com says it's slow)?

If you can catch it twice, what good is immunity? And if it mutates, how can immunity work? Some flus are coronaviruses and we're not immune to any flu viruses. That's why we need different vaccines every year.

[chris]

This interview we released last week about tracking coronavirus mutations will likely prove informative for you.

The bottom line is that all viruses mutate and this one is no exception. But, the rate, at a couple of mutations per month, is pretty slow.

[evan_au]

a couple of mutations per month

Another way of looking at this mutation rate is: "one viral mutation for every 2 or 3 people the virus passes through"
- The incubation period is around 5-7 days before symptoms
- People are infectious from perhaps day 4 to maybe day 14
- The time between one person getting it and them giving it to the next person is about 4-14 days
- So an infection chain has about 2-7 people per month
- There is one viral mutation for every 2 or 3 people the virus passes through
- This is equivalent to the quoted statement "a couple of mutations per month"
- However, if there are 100,000 people infected in a region, that represents 200,000 or 300,000 mutations per month...

With 30,000 RNA bases that could be misread...
- We could assume that maybe 50%-90% of these mutations would disadvantage the virus, and it would die out (at least, compared to its unmutated siblings)
- That still leaves >3,000 places where a mutation could occur
- And there are 3 possible changes in each position
- If the population of a region is well-mixed, tracing infections would be hard

With social distancing and travel restrictions to a very local area, if you have 10 contacts per day...
- The odds are that a mutation in one person won't overwrite a mutation to an "ancestor" virus
- So if we had high-capacity RNA sequencing of every infected person (including asymptomatic ones), it might be possible to track down who you caught the virus from.
- Of course, accurate RNA sequencing is still an expensive, slow and error-prone process

The GISAID group tracking influenza virus genomes quickly adapted their tools and website for the coronavirus, and now host a family tree of coronavirus too.
- They now have around 12,000 viral sequences from around the world (including the first ones released by China)
- They have lots of analysis by region; you can drill down into various views
- They commented that it did not take long for the strains in Europe to become very much mixed

World coronavirus genetics overview: https://www.gisaid.org/epiflu-applications/next-hcov-19-app/
Home page: https://www.gisaid.org/