The pandemic virus as evolution professor

Like it or not, the pandemic virus (SARS-CoV-2) is giving us all lessons in evolution and natural selection. The latest is one of the clearest examples of natural selection you are likely to see.  It is very clear cut, but to leave almost no one behind, I’m going to put in a lot of background material which will bore the cognoscenti — they can skip all this and go to the meat of the issue after the ****

The genetic code is read in groups of 3.  Imagine a language in which all words must be 3 letters long. 

The dog ate the fat cat who bit the toe off one mad rat.   Call this the reading frame, in which the words all make sense to you

Any combination of 3 letters means something to the machinery inside the cell responsible for reading the code, so deleting the f in fat 

gives us 

The dog ate the atc atw hob itt het oeo ffo nem adr at.   So this is a shift of 1 from the reading frame.  While it may not make sense to you, it makes sense to the cellular machinery. 

Now let’s delete 2 letters (in a row)

The dog ate the fat cat who bit the tof fon ema dra t.  

Not much sense after the deletion is there?  Or at least a completely different message.  This is a shift of 2 from the reading frame.

Now 3 letters (in a row)

The dog ate the fat cat who bit the toe off one mad rat.  

This gives 

The dog ate the fat cat who bit the tff one mad rat.  

Which has a funny looking word (tff), but leaves the rest of the 3 letter words intact (one mad rat).  This is called an in frame deletion. It basically lops out a single 3 letter word.  

Lopping out 4, 5, 6, .. letters will just give you one of the 3 patterns (frame shift of 1, frame shift of 2 or no frameshift at all) shown above (but nothing new)

*****

Now the business end of the pandemic virus is the spike protein, and these are where the mutations everyone is worried about occur.  The spike protein binds to another protein (ACE2) on the surface of human cells and then the virus enters causing havoc.  All the vaccines we have are against the spike protein. 

The spike protein is big (1,273 different 3 letter words).  

Mutations occur randomly.  We now have something called GISAID (Global Initiative on Sharing All Influenza Data) which has well over 100,000 genome sequences of the virus.  

Other things being equal we should see as many 1,  4 (3+1), 7 (2*[3] + 1), 10 letter deletions as 2, 5 (3 + 2), 8 ( 2*[3] + 2) , as 3, 6, 9, 12, letter   deletions.

The set  1, 4, 7, 10, . . represents a shift of 1 from the original reading frame, the set 2, 5, 8, 11 … represents a frame shift of two and 3, 6, 9 .. represents a set of deletions producing no frameshift at all.

Since thousands on thousands of experiments show that mutations occur randomly, 1/3 of all deletion mutations should show a frameshift of 1, 1/3 of all deletion mutations should have a frame shift of 2, and 1/3 of all deletion mutations should produce no frameshift at all. 

Well the authors of Science vol. 371 pp. 1139 – 1142 ’21  looked at 146,795 viral sequences and found 1,108 deletions in the gene for the spike protein.

They did not find each of the 3 types of deletions occuring to the same extent (1/3 of the time).  Among all deletions, 93% were in frame.  

Why? Because out of frame deletions change everything that comes after them. 

Recall

The dog ate the atc atw hob itt het oeo ffo nem adr at.  

This means that a functional spike protein won’t be formed, and the virus won’t infect our  cells, and it certainly won’t be found in GISAID.  

Ladies and Gentlemen you have just witnessed natural selection in action. 

Actually it’s even more complicated and even more impressive than that.  The in frame deletions occurred in one of four areas, which happen to be where antibodies to the spike protein bind.  So the out of frame deletions were selected against, and the in frame deletions were selected for. 

The blind watchmaker in action.

Another way to see how improbable it is that random choice should choose one of 3 equally probable possibilities 97% of the time, imagine that you are throwing dice.  You throw a single dye 100 times, and 97 times you get either of two numbers (say 3 and 6) .  You know the dye is loaded.  The load being natural selection in the case of genome deletions. 

 

 

 

 

 

Death rates from coronavirus drop in half 2 months after Georgia loosens lockdown restrictions

There were apocalyptic predictions of doom when Georgia loosened its lockdown restrictions against the pandemic coronavirus SARS-CoV-2 on 25 April.  Here they are

From The Atlantic — “Georgia’s Experiment in Human Sacrifice — The state is about to find out how many people need to lose their lives to shore up the economy.” — https://www.theatlantic.com/health/archive/2020/04/why-georgia-reopening-coronavirus-pandemic/610882/

 From the Washington Post — https://www.washingtonpost.com/opinions/2020/04/21/georgia-leads-race-become-americas-no-1-death-destination/

From MSNBC — https://www.msnbc.com/rachel-maddow-show/georgia-s-kemp-neglected-warn-people-about-his-dangerous-gamble-n1189826

A month later (25 May) not much had happened —

7 day moving average of new cases of COVID19 ending 25 April — 740

7 day moving average of new cases of COVID19 ending 13 May — 525 (the state allows 14 days for all the data to roll in, so the last date they regard as having secure numbers is the 7th of May and here the number is 539)

7 day moving averages of deaths from COVID19 ending 25 April — 35

7 day moving average of deaths from COVID19 ending 13 May — 24 (the state allows 14 days for all the data to roll in, so the last date they regard as having secure numbers is the 7th of May and here the number is 27).

Back on 25 May I wrote “People who assumed (on purely correlative evidence) that lockdowns prevented new cases, and that lifting them would cause a marked increase in new cases and deaths, are clearly wrong.  It’s possible that cases will spike in the future proving them right, but pretty unlikely.  It’s only fair to give the doomsayers a sporting chance and followup is planned in a month.”

So here’s the followup.   The 7 day moving average of daily deaths had dropped to 17 as of 11 June.  Remember Georgia waits 14 days as data filters in to regard the numbers as definitive.  Here’s the link — https://dph.georgia.gov/covid-19-daily-status-report

So the death rate from COVID-19 dropped in half 2 months after Georgia loosened some of the lockdown restrictions.

There are only two useful statistics in all of this.  The moving average of the daily death rate and the number of COVID19 cases in the hospital.  I no longer follow the number of new cases, because they include people with a positive antibody test (all of whom have recovered).  We know that most cases are asymptomatic.  It’s very hard to get the second number of people sick in the hospital with COVID19 (I’ve tried with no luck).  COVID19 used to mean that you were sick — no longer, it now counts positive antibody tests, rendering the number relatively useless.  By choosing who to test, numbers can be easily inflated — https://luysii.wordpress.com/2020/06/22/new-york-city-covid-19-cases-spiked-today-stock-market-futures-tank/

Daily death rates are great for cherry picking scare headlines — it’s worth looking at this article from Tampa — https://www.wtsp.com/article/news/health/coronavirus/florida-coronavirus-cases-hospitalizations-deaths/67-4bbd0c35-6742-4f51-a59a-ea1d101f54ea

It contains a great figure with the number of deaths each day from March onward on which is superimposed the moving average — the range is from 10 to 100.  Even more impressive is the fall on weekends and the rise during the week.

Fortunately, every Friday  Florida releases the weekly results for antibody testing, so we’ll be able to see how many of these new cases of COVID19 are people who have recovered from it.

Here’s another link — well worth looking at — with the number of new cases in Florida in one graph (with the marked increase in the past week) and the number of death from the disease just below.  The deaths in the past week are the lowest they’ve been in a month — https://experience.arcgis.com/experience/96dd742462124fa0b38ddedb9b25e429

A Tale of Two States (with apologies to Dickens), the denouemont

Two days I posted the following puzzle — here is the answer and a bit more

A friend in med school, a classic University of Chicago graduate, was fond of saying “that’s how it works in practice, but how does it work in theory?”

Well, this country is currently in the midst of an immense social experiment (lockdowns) essentially based on theory (models).

We’re about to find out how it worked in practice.

Here are some recent statistics from two states.

State 1

3 day moving average of new cases of COVID19 ending 25 April — 2778

3 day moving average of new cases of COVID19 ending 13 May — 901

3 day moving average of daily deaths from COVID19 ending 25 April — 177

3 day moving average of daily deaths from COVID19 ending 13 May9 — 96

 

State 2

7 day moving average of new cases of COVID19 ending 25 April — 740

7 day moving average of new cases of COVID19 ending 13 May — 525 (the state allows 14 days for all the data to roll in, so the last date they regard as having secure numbers is the 7th of May and here the number is 539)

7 day moving averages of deaths from COVID19 ending 25 April — 35

7 day moving average of deaths from COVID19 ending 13 May — 24 (the state allows 14 days for all the data to roll in, so the last date they regard as having secure numbers is the 7th of May and here the number is 27).

One state loosened its lockdown restrictions 25 April, the other had them in effect through 13 May.  Your job is to figure out which one did and which one didn’t.

The denouement — State 1 is Massachusetts (which kept the lockdown) and State 2 is Georgia which loosened them on the 25th of April.

As usual, actual data answers some questions but raises new ones.  Contrary to the disasters predicted (see later), in Georgia the new cases of symptomatic pandemic flu declined by 29% and the number of deaths declined by 22%.  The 13th of May is way past the longest possible incubation period for cases beginning prior to 1 May.  So from this, the conclusion one might draw is that the lockdown was ineffective.

But hold on. Massachusetts also showed declines in new cases and deaths, and by greater amounts 68% and 46% than Georgia (29 and 22%)  implying the lockdown was of some use (in accelerating the decline in cases and death).

Pandemics and epidemics have a natural history of peak and decline, in the USA our pandemic is on the decline.

People who assumed (on purely correlative evidence) that lockdowns prevented new cases, and that lifting them would cause a marked increase in new cases and deaths, are clearly wrong.  It’s possible that cases will spike in the future proving them right, but pretty unlikely.  It’s only fair to give the doomsayers a sporting chance and followup is planned in a month.

Here are a few predictions of doom.  Future predictions and definitive statements from these sources should be taken with a grain or more of salt.

From The Atlantic — “Georgia’s Experiment in Human Sacrifice — The state is about to find out how many people need to lose their lives to shore up the economy.” — https://www.theatlantic.com/health/archive/2020/04/why-georgia-reopening-coronavirus-pandemic/610882/

 From the Washington Post — https://www.washingtonpost.com/opinions/2020/04/21/georgia-leads-race-become-americas-no-1-death-destination/

From MSNBC — https://www.msnbc.com/rachel-maddow-show/georgia-s-kemp-neglected-warn-people-about-his-dangerous-gamble-n1189826

Georgia is the canary in the coal mine

The decision of the Georgia governor to relax some restrictions on activity and commerce 25 April was not met with universal acclaim.  In fact here’s how an article in the usually rather stolid The Atlantic puts it

       “Georgia’s Experiment in Human Sacrifice

The state is about to find out how many people need to lose their lives to shore up the economy.”

Sounds like a backwoods preacher calling forth hellfire and damnation on those not following the will of the Lord (experts in today’s parlance), doesn’t it?    Here’s the link — https://www.theatlantic.com/health/archive/2020/04/why-georgia-reopening-coronavirus-pandemic/610882/

Well, we’re about to find out if they’re right.

Here is the link to the very excellent website of the Georgia Department of Public Health –https://dph.georgia.gov/covid-19-daily-status-report.  The site is updated 3 times daily.

The site is extremely conservative in the way it reports the data.  Realizing that data doesn’t always flow in to the Department promptly, the number of new clinical cases of illness (COVID19) or the number of deaths is recorded as secure only after 14 days have passed.  So today being the 17th, they regard the number of cases on 3 May as secure.  The rest of the numbers on the chart (which go right up to today) are plotted on the chart as well. And you have the option of looking at the data three different ways (cumulative cases, deaths, daily new cases)

Daily new cases, and deaths  appear to have been stable from the 25th of April to 3 May.  However depending on who you read, the incubation period of symptomatic pandemic flu (COVID19) ranges from 2 to 11 days (longer times have also been suggested). If there is going to be a spike in cases we should see it in the next week or so.

It is also far too early to say that the cries of doom from the Atlantic are wrong or that the people pushing for relaxation are right, but we certainly haven’t seen a spike in the daily number of new cases or deaths so far.

Stay tuned

Addendum 18 May — of course the above admonitions about jumping the gun didn’t stop the following right wing site from claiming victory today — here is a link — https://spectator.org/you-only-die-once/ — however, it might be worth reading for links and quotes to other cries of doom from MSNBC, Washington Post and the Associated Press

The presidential election will be decided in the next month — revision for clarity

Apologies to all for the previous post which was far murkier than it could have been.  The problem was that a ‘case’ of the pandemic coronavirus can mean 3 very different things.  These distinctions are tedious but crucial.

Meaning #1 — COVID19 — People who are clinically ill with the virus (official name SARS-CoV-2).  These are the people that  may die of the illness, although most do not.

Meaning #2 — The viral genome has been found in your saliva.

Meaning #3 — You have antibodies to the virus in your blood.

Here is the distinction between #2 and #3 — Antibodies (proteins) and genomes (RNA) are completely different chemically. Finding the actual genome (RNA in this case) of a virus in an individual  is like seeing a real bear up close and personal.  This can do you some damage.  In contrast, antibodies to the virus are made by an individual who has been infected by the virus in the past.       Antibodies are like seeing the tracks of the bear without the bear itself. You can’t see tracks without the bear having been present at some point in the past.  Antibodies mean you were infected at some point whether you knew it or not.

OK, so here’s another shot at what was I saying in the previous post.

I find it very sad that loosening the restrictions on activity has become so political. The left says that it will be a disaster and that cases and deaths will spike (meaning #1). As far as I’ve seen, they never say they hope they’re wrong.  The right says that deaths will continue, but the rate won’t increase.  There is evidence for both sides, but in the coming months we’ll actually have data one way or the other.

One thing is certain.  The number of cases of positive viral culture (meaning #2) will increase.  It has to because more people will be tested. So far, we’ve only studied around 1/1,000 of the population.  No one has ever said the lockdown will prevent new infection.  It hasn’t, but it has slowed things down.

I’m hoping that cases of COVID19 and death will not explode.  Not because I want Trump to win, but because getting people back to work  would be good for the country.  Should that happen, the anger of those who lost their jobs or businesses during the shutdown will be formidable.  Trump will win.

Should deaths from COVID19 explode (meaning #1) as restrictions are lifted, Trump is toast.

We should also get some idea of the percentage of the population who have been infected (manifest by antibodies to the pandemic virus meaning #3).  It almost certainly will increase, unless those already showing the antibodies lose them (which is unheard of happening this fast inFor  the antibodies we’ve studied in the past).

I’m cautiously optimistic that not much will happen when restrictions are eased. Here’s why.  All the studies on antibodies (meaning #3) done so far show they are 10 – 100 times more prevalent than cases where the virus is cultured (meaning #2).  For example 20% of Manhattan sampled population have the antibodies.  This implies that most infections with the pandemic coronavirus are asymptomatic.  

Another viral disease with a high prevalence of antibodies is infectious mononucleosis.  90% of adults in the USA have antibodies to mono, but far fewer than 90% were ever sick.

So the number of cases with positive culture (meaning #2)  isn’t what’s important.  It’s how many of them get sick with COVID19 (meaning # 1).  I think we have very good past statistics on the number of deaths and cases of COVID19   It will be clear if there is a spike in COVID19.

However be careful not to read too much into the first week’s statistics after restrictions are lifted, as there is a lag period of 2 – 11 days between infection and clinical illness.  Also try to understand which of the 3 meanings of “case” the article you are reading is talking about — this won’t always be possible.