Comment by rleigh | Hacker News Reader

Comment by rleigh | Hacker News Reader

Parent Full thread

Evolution occurs gradually, one base-pair at a time through replication error (or various other effects). Occasionally you get big changes through duplication, transposition or deletion of whole parts of a sequence. But the latter hasn't occurred here--we're looking only at the former. When we study evolution over time, we can look at the accumulation of silent and non-silent point mutations. And we can also pick up any larger insertions and deletions. From this data we can create a graph of the changes over time based upon relatedness (the phylogenetic tree). This is the data shown in the article.

When looking at a large sampling of data across an entire population, we would expect to see a gradual accumulation of point mutations over time. And very rarely, some larger insertion or deletion event. None of this is new. We had all the tools to do this for over two decades. I was using Perl for bioinformatics 15 years back.

So given the assumption that we would see a gradual accumulation of point mutations over time, if we see any big changes appear out of the blue, that's a flag to investigate more closely at what happened. That's not "creationism" of any sort. Given that the point mutations give a very clear fingerprint, we can immediately find the most closely-related genome, which is why it was found to match a sample collected in early 2020. That completely breaks our expectations of it missing a whole 12 months of natural accumulation of point mutations. We might not sample perfectly--we might occasionally see two or even three point changes because we failed to sample those intermediate states. But to have a jump spanning dozens of changes--that's just not matching our existing expectations, particularly when we tracked nearly every Delta change.

Now that might be real. But the probability of that being natural is very, very low. Almost nonexistent. But the chances that this was stuck in a freezer and pulled out, skipping over those 12 months of mutation-gathering while frozen in a tube, are unfortunately a much more likely possibility.

But we never have the full picture. 2020 was a year where several countries closed their borders and restricted domestic travel. Many countries were not sequencing the virus. It's not that far fetched that we'd see a descendent of an ancient variant if it had been in hiding in some remote isolated village in a poor country.

While what you say is certainly possible, it's not (IMHO) very likely. It doesn't matter that countries closed borders and restricted travel, or that not all countries were sequencing with the same rigour. That doesn't affect the eventual outcome as we randomly sample patients in the rest of the world.

If a country were hypothetically completely isolated and not sequencing, then we could have a new unknown variant develop. But from a given previously known origin genome, it would be mutating within that population over time at a fairly predictable rate, which would lead to a family of variants rooted at that known origin. We would see a spectrum of variants in a cluster branching from the origin. However, it clusters far from its origin, and this does raise a red flag.

It's possible that as we collect more data, we will "fill in the gaps" and get a better picture of how it evolved to the point it is at today. If there is a hypothetical country where this evolved naturally, those variants will still exist within the local population, and we will gradually sample a range of them in other countries over time as they too travel abroad where they can be detected.

If however it is not natural, then those intermediates will not exist, or will only exist as frozen down passages in a lab freezer. Maybe we'll identify the lab in time, if that's the case. They should be able to match it if they sequenced it.

Whichever hypothesis is correct, the data can't lie. We will know one way or the other fairly definitively in time.

You are confusing our lack of knowldege of the intermediate steps with the lack of interim steps.

They might simply have been lost in the crowd or not even noticed, depending on where they were posted. How carefully have we studied the variants in animals?

No, I'm not doing that. We might not yet have sampled the intermediate steps, if they exist, but it's inevitable that we will do so if it has a natural origin. Given the phylogenetic distance between the omicron cluster and its root in the phylogenetic tree, there should be several hundred to thousand intermediates if it's naturally occurring, of which we should be sampling many dozens (modulo selective pressure and sample availability biasing the sampling). Assuming that mutations occur randomly and with equal probability, then you would naively expect if the distance is between the cluster and the root is n base pair changes, then there would be n raised to the fourth power variants (assuming equal selective pressure). The reality is a bit more subtle than that, but I hope it illustrates the general scope and scale of what has taken place here. Given that scope and scale, the complete lack of any sequence data to date is astonishing. Particularly when we do see all those intermediate states for the other variants. Does that make sense?

As for variants in animals, that's a possibility. But it's more likely to be a red herring. Domestic animals will be infected with the same strain as their owners, and any reinfection would be very likely to be similar to the original. Transfer to a wild animal reservoir and back into the human population is so rare and unlikely that it is barely credible. Infection of humans from wild animal reservoirs is a rare event. To have it happen both ways within 12 months is not credible.

I could be wrong. This will only be answered by collecting more data to improve our understanding.

Why should it be inevitable? How long do non dominant variants survive inside and outside a host? We don't even know in which country omicron originated even less who was patient zero.

We could easily miss whole strains of the virus but because non of them lead to a new dominant variant we never know the even existed.

It's inevitable because even though one variant may become dominant, there will still be isolated pockets of the other variants. It's very unlikely that these variants will be entirely eliminated across the whole globe. Look at the epidemiology of HIV and the spatial distribution of variants as a point of comparison. Likewise for other viruses. Even if these other variants continue to evolve from the initial root, they will still be phylogenetically distinct from the current omicron clustering, somewhere intermediate between the root and existing omicron cluster. If it's out there in the wild, it will be picked up by PCR eventually. If it's not out in the wild, that big gap will remain, and a lab somewhere has questions to answer...

The starting point for the omicron cluster would be an interesting bit of information, but it doesn't necessarily have any bearing on where the intermediate strains are located. But none of that matters. If it's out there, it will eventually be picked up by PCR. We can use phylogenetic analysis to retrospectively reconstruct the evolutionary tree of every organism we have sequenced to date. Doing it just for the coronavirus variants is not hard (I have done it in the past including bootstrapping). Whether it's a month, a year or a decade from now, we will be able to answer the question by reconstructing the tree.

This is really basic bioinformatic analysis, in routine use for over two decades. None of this stuff is remotely new or groundbreaking.

The other questions you are asking are not really relevant to the problem. The virus does not survive long outside a host; it has no bearing upon the problem at all. The same applies to inside the host. All variants will continue to infect and propagate themselves to varying degrees within the host populations. It's endemic at this point.

One other point to bear in mind is that the mutation rate of the polymerase is effectively a constant, in terms of mutations per n kilobases. So we would expect all variants to accumulate a certain number of mutations averaged over time--the sequence will not remain static since the virus has to continually be passaged through mammalian epithelial cells to maintain and propagate itself. There will be some variation, but if it's been frozen down and manually propagated, it may have a dramatically different number of mutations compared with the other contemporary strains that have spent most of their time in living hosts.

Reply on news.ycombinator.com