@tgof137: In that case, they spelled it:...

In that case, they spelled it:

CGG ATC AGG CGC

So they created 3 R's, and only one of them was CGG. The two adjacent R's were AGG CGC

Scientists have several programs that they use, to recommend codons.

None of those programs would have recommended CGG CGG in this spot.

I think I first heard the CGG CGG claim when Yuri Deigin went on Bret Weinstein's show.

It turns out Yuri isn't even the first guy to come up with this idea.

When he first looked at SARS-CoV-2 he wrote a long blog post on why he thinks it's engineered.
yurideigin.medium.com/lab-made-cov2-…

In that blog post, he wrote a whole section on “codon usage”, where he carefully compared SARS2 with RATG13, SARS1, and several other viruses. Despite all that work, he never even noticed the double CGG.

It looks like he first learned about it from some random guy on Twitter, on May 2020.

So, this was never an obvious part of the lab leak theory, and it hasn't been embraced by all lab leak theorists.

It's more like:
people have just tried creating lots of different versions of the lab leak theory and it evolves to include the most popular claims.

The CGG CGG theory has grown to be one of the most popular claims.

Since there's no evidence that engineers actually prefer CGG codons, you could have used other letters to make this theory.

Notice that CGA and CGC are also rare in bat viruses and more common in human DNA.

If the sequence had been CGA CGA or CGC CGC, Yuri could have just as easily said that those are the codons that engineers prefer -- the claim would be made up, either way.

Suppose that you live in some state where license plates usually have 3 letters and 3 numbers.

If you see a license plate where the numbers are 000, that's unlikely, the odds are 1 in 1,000 of finding that, just like the odds are 1 in 1,000 of CGG CGG.

If you also knew that scientists always have license plates that say 000, then that might be good evidence that car was driven by a scientist.

But if scientists don't actually have special license plates, then this is not actually meaningful.

000 is unlikely, but so is 111 or 222 or 333 or 459. Each of those numbers is 1 in 1,000.

You could see some random number and tell people, "that's the license plate scientists always have".

But that's meaningless, because you could have pretended that's true for any number.

That's basically what Yuri Deigin and Steven Quay did, with the CGG codons theory.

Okay, so the theory that scientists love using double CGG codons is made up.

But I still haven't explained why SARS-CoV-2 has rare codons here, if Covid is natural.

Science doesn't have a perfect answer for that -- if the virus evolved naturally, no one consciously made that choice, so we can't say what nature was thinking at the time.

But we do have some guesses for why it's there.

One experiment tried mutating away the CGG codons.
pmc.ncbi.nlm.nih.gov/articles/PMC82…

(from CGG CGG to AGG AGG)

They found that the updated virus was less infectious than the one using CGG.

The issue had to do with protein folding. Their updated version made more spike protein, but the proteins
folded less accurately.

The CGG codons slow down the protein translation, and that improves folding accuracy.

That may be the reason why evolution favors this.

If the CGG CGG was unnatural, or nature selected against it, then it would mutate into some other spelling of RR, over time.

You can change only one letter of CGG in 4 different ways and still get R. It could change to CGA, CGT, CGC, or AGG.

We’ve now had millions of covid cases around the world.

As of mid 2021, after 18 months of evolution, CGG CGG was still found in 99.85% of them.

As of 2025, it was found in 99.6%.

It’s not mutating away from that.

For whatever reason, that spelling works best.

Another thing we've learned, as the pandemic has progressed, is that Covid sometimes naturally gets insertions in random places in the genome.

Some of those random insertions have a double CGG.

These insertions come from errors in the virus' replication, where some strand of RNA gets copied in to the virus. That RNA can come from some other strand of the covid virus, or another virus.

But the insertion can also come from some strand of host RNA, in which case it would have the human frequencies for CGG (> 20%) and then finding a double CGG is not that strange.

The same thing could have happened when Covid was evolving in a bat, where CGG is 20% or in another animal like a raccoon dog.

Now that we know that double CGG is not favored by scientists and can happen in nature, we can revisit the odds.

At first, we were thinking: this is 100% likely to happen in a lab but it has 1 in 1,000 odds in nature, therefore lab leak is 1,000 times more likely.

Now, we might be thinking it's 1 in 36 odds that a lab would do this (if scientists are just picking randomly, they have 6*6 = 36 choices for RR).

An analysis by Guy Gadboit found it's actually 1 in 200 odds to get the double CGG if it's copied from another bat virus:


So the ratio there is only 5.5X in favor of engineering (5.5 = 200 / 36)

View Tweet

Another analysis by Guy Gadboit found that, if the insertion came from host RNA, the ratio is only about 2X in favor of engineering:

View Tweet

Next up, remember the result described above where CGG CGG is actually the optimal choice here.

In that case, evolution would gradually turn any other sequence into CGG CGG.

View Tweet

If that's true, that means the odds of finding it for a natural virus could be close to 100% while for an engineered virus, it might be 1 in 36 (or lower).

Now the ratio points strongly in the opposite direction -- this feature is 36X more likely in a natural virus than in an engineered virus!

The exact odds here are hard to pin down because:

We don't know if CGG CGG is evolutionarily perfect or if it's just "good enough"

We don't know how long the virus was evolving before it was discovered

We don't know if the insertion came from another virus or from the host

In any case, we can say that the odds here do not strongly favor engineering.

The odds could be anywhere between weakly in favor of a lab (5.5X) or strongly in favor of nature (36X)

And the main reason that this ever sounded like a strong theory is because someone lied and pretended that scientists prefer double CGG codons.

Again, I thought the CGG CGG argument sounded meaningful when I first heard it on a podcast.

Maybe you did, too.

But it turns out that most podcasters aren't scientists, and many of them are not honest or competent at fact checking the claims they make.