I can't say I'm terribly well versed on the conspiracy theories, but from I gather they fall into two classes: 1) The virus escaped from a laboratory that was doing legitimate research. 2) The virus was deliberately engineered and released as a weapon. The facts don't support the first scenario, but at least it's internally consistent with the false claim of having sequences spliced in from other viruses. I'll come back to why people are saying there are insertions from HIV and why that's incorrect. Another argument against this is that methods for modifying a genome require also inserting other well known sequences needed by the insertion machinery. These sequences haven't been observed. It's possible they could have been removed, but that wouldn't be done in a research setting, which brings me to the second class of theories. In scenario 2, as you point out, you wouldn't want obvious insertions, yet that still gets brought up and people seem to jump back and forth between the two scenarios. It's hard to prove a negative that you couldn't engineer a virus to look natural, but in my earlier post I was trying to argue it seems quite difficult. You could make it better at binding to surface receptors on lung cells, but there are other critical features like the long incubation time and ability to spread through seemingly healthy carriers. I don't know how you would begin to engineer for that short of a long process of large scale human testing. I don't you'll see studies like that anytime soon. For one thing, it's very difficult to know whether a mutation is functional or not. Our ability to interpret the genome is quite limited compared to reading a computer program. It really requires a painstaking process of making modifications and seeing what the result is. It's quite common that even when you know a sequence must be important (because of high conservation across species), you still don't see an effect when you delete it, presumably because you aren't observing it in the right environment. For another thing, there are limited resources to explore a lot of directions right now. After some cursory work to shoot down one shoddy paper, researchers will focus their attention on directions more likely to be fruitful rather than making sure every nail is in the coffin. The claim about HIV comes from this paper, which has been retracted. The impression it gives is that compared to SARS the new coronavirus contains four well-defined insertions which all come from a single other virus: HIV. Intuitively that would seem to be quite unlikely by chance. In reality, there are nearly 300,000 different versions of the HIV envelope glycoprotein that are in the sequence database. Here's a comparison of the first two to give you an idea of how much they differ. The four insertions are not all found in a single strain, and the insertions are quite short, so the odds of them occurring in one strain by chance are high. Doing the math properly is actually quite tricky due to selection pressure and the fact that some amino acids are relatively interchangeable, but there are many billions of virus particles produced per host, so the search space can be explored pretty well and it's relatively easy to get the same short sequence evolving independently if it confers a selective advantage. There aren't actually four well-defined insertions. I've attached Figure 2 from the paper showing the four insertions. Why are the last four residues included in insertion 2 but not any more of the residues to the right? Why is the insertion to the left of insertion 3 ignored? Because if they defined the insertions differently they wouldn't match HIV anymore. Insertion 1 is GTNGTKR (with several mismatched residues to the left), but only TNGTKR matches HIV. It's pretty clear they are retroactively defining the insertions as whatever matches HIV to make a nice story. I'll try to get to some of the points in the youtube video later this evening.
Thank you, this was very helpful. I was aware that the HIV paper was retracted, but apparently they were supposed to reprint it. That being said, your explanation makes sense. As a layperson, I wonder if there is software to simulate or try to predict (through machine learning) the impact of a change like you are suggesting to shortcut the laborious process of change-build-test-change-build-test. And if such a thing exists in a basic form publicly, whether it is a competitive or tactical advantage for engineering bioweapons. For example, NSA was able to backdoor elliptical curve encryption by releasing a specific set of random numbers that, if used, would make it feasible to crack your messages. This wasn't known for years and was a tactical advantage.
It's an active area. Some tools are PolyPhen and SNPeffect. However, everything I'm aware of is focused on determining if a mutation is likely to render a protein nonfunctional. For example you get your genome sequenced and want to know which genes are disrupted and are they known to be involved in cancer or other disease pathways. It's relatively simple because you're just looking at whether the substituted amino acid is similar to the original and is it in an important part of the protein, but even those tools aren't that great. Trying to predict a new function is unimaginably far away at this point.
No need for that. Congress is already doing the job. Pelosi and McConnell can´t agree on the $ measures. They still think this is a clown show...