Previous post was taken down within a few minutes by an unhelpful poster causing it getting it flagged with an abusive post. Two good questions were asked there so I will repost them here:
>Posted by 'version_five'
"This sounds like one of those "bible code" things.
As someone who knows nothing about DNA sequences, what is the probability that two codes could be matched? (In a birthday paradox sense where you can go looking anywhere)"
>Posted by 'whoomp12342'
"isn't it well known that one of the main reasons why we got a vaccine so quickly was because we had done a ton of research when SARS originally broke out?"
Maybe that is the case, but they did not flag it, they posted a message with several violent keywords and the post was immediately flagged. It seemed different from normal flagged posts because all of the reply buttons disappeared and as I hit reply on a comment I was in the process of writing, I got the message "You can no longer reply in this post" or something to that effect. I looked at other currently flagged posts and this seemed atypical. You can still reply within the flagged posts comment section. I assumed it was some sort of auto moderation action.
The reply buttons likely disappeared because it was also made "dead". Looking it up in your profile it shows "[flagged] [dead]". As to why it was made dead, I do not know.
>This sounds like one of those "bible code" things. As someone who knows nothing about DNA sequences, what is the probability that two codes could be matched? (In a birthday paradox sense where you can go looking anywhere)
The genome in question is about 30000 bases long, which isn't nearly long enough for the outcome space size issue to come into play. Assuming independence, the probability is still very remote, effectively zero.
However, there is in no way any expectation of independence. SARS-CoV-2 is obviously closely related to SARS-CoV-1, which is obviously closely related to 2015/2016 ongoing vaccine research. (And perhaps gain of function research.)
https://news.ycombinator.com/item?id=29938340 was flagged by other users—quite a few of them. We can only guess why users flag things, but I doubt that it had anything to do with that trollish post. We've banned the latter account now.
To clarify this sequence has not been identified in any other virus in the wild. Additionally it codes for the section pertaining to the furin cleavage site which does not exist in any other observed beta coronavirus.
You are correct. I should have stated furin cleavage sites have not been observed in another beta coronavirus. (Cold, MERS, SARS) I corrected my statement. Thank you for pointing that out.
I think the point is that the sequence is searchable within the database of known viruses (virii?) and it is not speculation, it literally does not exist in any sequenced virus in the wild. I'm not sure what you mean with the goalposts comment.
And by "goalposts" I mean that if you recall, the initial allegation was that COVID was a bioweapon developed by the Chinese. Then it was negligently leaked from a lab doing gain of function research funded by Obama. Then it was just leaked. People seem to want to believe this theory regardless of when facts get in the way. A year ago, there wasn't much evidence against it. Now there are animal finds in Laos that make some of the connection. Not proof, but positive evidence. From the field.[1] Not from cherry-picked database searches.
At some point, you have to ask yourself with Ockham's razor: if it really is this simple, then why isn't it widely accepted? Has China gotten to everyone? If so, then why can I post this to my substack? Sometimes the expert opinion is right. I guess sheeple like me will never know. Except in reality, I'm quite open to this hypothesis, I am already a China hawk so I need no motivation there. I just don't think there's as much evidence as people who want to believe this think there is.
For what it's worth, using the model of radius->radii for virus->*virii makes sense until you realize radius has an extra 'i' already. Worse, it's a second declension neuter despite the -us and so should be vira just like modicum->modica, whereas radius is second declension masculine.
It's not the responsibility of people with domain knowledge to go around and debunk every substack post. But to believe this post you have to believe that no one else discovered this, or, of they did, they are suppressing it.
Which is comical. Most scientists would rush to publish a new finding and take credit. There's no "establishment" conspiracy to suppress evidence of the lab leak theory. The Biden administration ordered further investigation as one of their first acts. No suppression there. So, who then? The WHO is being too nice to China? I agree! Doesn't prove anything.
If it's as easy as the OP says, we would know.
This is a false conjecture. Chinese scientists experienced a clamp down on virology publishing around the original time of the outbreak, and at this point there is social stigma attached to lab leak that may very well encourage people not to publish.
Nevermind the diplomatic implications. If, hypothetically, tomorrow the official stance of the U.S. Government changed to hold that China was responsible for the manufacture of the virus, that despite the part the U.S. had in funding it being facilitated by failures in process at NIH, this would signal a sizable shift in attitude toward the Chinese, and could very well escalate tensions to a degree no one has a stomach for.
>There's no "establishment" conspiracy to suppress evidence of the lab leak theory. The Biden administration ordered further investigation as one of their first acts. No suppression there.
You have no idea how much classified data and information may exist. There is a whole corpus of knowledge kept out of the public's view for years because of how the Executive feels about how much more complicated it would make their lives. I would not dismiss official perception management as quickly as you do. The government does not hint. It does not offer, and leaves it to the public to do their own research until such time as the public largely figures it out independently; and then still will not confirm or deny. Look into how the body of knowledge around nuclear weapons was cobbled together by uncleared or non-government affiliated researchers. FOGBANK was such a closely guarded secret, it had to have it's manufacturing process reserved. Same with various B-1 part manufacturing processes.
I have no idea how old you are, or how much time you spend observing how groups of people work and behave emotionally, but it's fairly obvious to me that this type of finding is something no one with an appreciation for second order effects wants to let out of the bottle without a lot of thought, or blind faith it is the RIGHT thing to do; least of all those virologists who either truly believe GoF research is necessary, or those that wish to conduct research without a billion hoops to jump through. Further, not all the old-timers from WWII are dead yet, and those with parents who lived through the Cold War are fairly certain that if it comes down to a WWIII, there may not be a return to any semblance of normality in the aftermath. This is exactly the kind of flashpoint in history that resonates with lighting that kind of fuse.
It seems you might be lumping me in with some other claims that I have not made. I agree that it is wise to stick with the facts and keep speculation to a minimum. It is a shame that this topic has become so politicized because we deserve answers. We cannot prove what did happen, but through careful observation we maybe can eliminate the impossible.
I don't know much about this particular topic at all -- but your Ockham's razor point is a really bad one; history is replete with instances of "very simple and also correct answers/solutions/theories" that are considered quackery or wrong at the time.
If you believe in all the conspiracy theories, you're a sucker, but if you believe there are ZERO conspiracy theories, you're also a sucker.
Please omit swipes like that last sentence from your posts to HN. That sort of cheap flamebait leads to tedious flamewars, which we want to avoid here.
Everyone always feels like the other person started it and did worse. That's a universal cognitive bias. Also, when people get moderated, they often instincitvely feel like the mods must be against them or their views. So I understand the temptation to react in these ways, but it doesn't help make the site better.
You (by which I mean not you personally, but everyone here) need to follow the rules regardless of what other people are doing. Taking the intended spirit of the site more to heart would be helpful: https://news.ycombinator.com/newsguidelines.html
Just to be clear, this doesn't imply that moderna released the viruses. Patents are publically visible, after all. It suggests that someone was told to insert a furin cleavage site, then went and "stack overflow copy-pasted" the moderna furin cleavage site, and this became the coronavirus. If we believe that this is what happened, it possible that whoever did this also looked up and copy pasted other furin cleavage sites, and thus the purely random odds of the eye-popping "moderna" aspect of the coincidence would considerably less (let's say, 1 in 100 or less, depending on how much work you think the hypothetical postdoc bothered to do). Do note that the moderna furin cleavage site itself is probably subjected to industrial selection pressures that enhance its delivery, which might also correlate to infectivity, so a non-uniform distribution of infectivity among stack-overflow-ed furin cleavage sites, that favors the infectivity of something like the moderna sequence, should be considered.
Given that this line is from a cell culture, wouldn’t the simpler explanation be that the virus has this sequence because it was replicated in that specific Moderna culture lineage?
This would be a particularly useful discovery, because it would dramatically limit the candidates for who did the gain of function research to organizations licensed this Moderna lineage.
Impressive work, but author's logic is unsound because it silently assumes the library or database BLAST uses is complete, yet it likely represents less than 1% of all sequences in the wild. The database BLAST uses only contains sequences that have been found and reported. If every person on the planet was a genetic expert with all necessary resources to find and identify new sequences, in 1000 years we still wouldn't have found all of them. The sequence in question that only matches HIV-1 may very well match a trillion^2 more sequences not in BLAST's database. Anything like this that is engineered likely has a mountain of documentation. Documentary evidence, if found, is a smoking gun in a way the furin cleavage site sequence can't be.
It would be interesting to me how likely it is to actually find random sequences in the database. Aka how much correlation there is between natural viruses. In my imagination it is plausible that if you take a particular virus that you will be able to find most of its subsequences (with minor modifications) in other viruses. In that case I think that his analysis is sound. On the other hand, if the other viruses appear to be randomly sampled from the set of all possible sequences than his reasoning does not cut it.
Genomes are sequenced when they're both accessible and are of scientific/economic interest. This results in an overrepresentation of:
- Humans and their pathogens
- Economically important organisms (e.g. crops and farm animals) and their pathogens
- Model organisms
For example, compare the number of Genebank sequences for HIV-1 [1] versus Feline immunodeficiency virus [2]. HIV-1 is additionally problematic because it has an extremely high mutation rate [3], making it more likely for a random sequence to match some sequenced HIV-1 genome.
The HIV virus has been sequenced 10's of thousands of times, and all those sequences are in the database that was searched. Likewise, there are thousands of copies of the E. coli proteome in the sequence databases. The nr database, part of which was searched, is extremely redundant (despite the fact that 'nr' stands for "non-redundant" -- no two identical sequences).
So unless you go to a lot of trouble (e.g. the NCBI landmark database), most sequence databases are a mixture of 10's to 10,000's of very closely related sequences (not a random sample), from a non-random sample of organisms (we tend to sequence pathogens).
Protein and DNA sequence databases are very large, but they are not random.
This is a semi-interesting piece (because of how poorly written it is and how hilariously confident the writer is), but it is embarrassing that the person writing claims to have a PhD in at least something semi-related to molecular biology, yet is trying to convince people that we can use independent probability when dealing with protein amino acid sequences. Certain sequences are seen time and time again in nature, just like certain traits visible to the naked eye have evolved independently. More importantly, this author dismisses natural mixing of SARS and HIV1 as "ridiculous" and provides no other explanation. It is possible that SARS-CoV-2 was created in a lab, but it is also possible that HIV-1 and SARS-CoV-2 co-infected a host.
Edit: to elaborate, certain amino acid sequences are seen again and again, and others are practically impossible to see in nature because amino-acids fold into 3D proteins...and each amino acid is like a Lego piece. There is a lot more detail needed to understand fully how tertiary and 4try protein structures form...but it is also possible to understand it decently well after an undergraduate degree in biology, chemistry, biochemistry, etc. A PhD in the field knows this inside and out. If I had to guess, the author of this post has a PhD in something like statistics or computer science, and thinks that they can apply high school level math to two fields (molecular biology and biochemistry) that they do not even have a high school level of understanding of.
To make a very stretched analogy (I am a doctoral student in the life sciences and a hobby programmer), this blog post is like saying that some Java source code is stolen because four of the class names are the same between two projects, and then using the number of possible characters in the class names and the length of the name to run some statistics. The problem being that no one has class names like "AeNOQ92bA"...in fact the vast majority of 8 character sequences will never be class names. Just like the vast majority of amino acid sequences (likely) do not exist in nature. And then you dig deeper and find out the class name is something like "MainDashboardSupervisorTree" (I do not know Java so forgive me). Then you also find out that the author of both classes...is the same guy who moved companies and likes particular naming convection, but never meant to copy stuff word for word. Similar to how SARS-CoV-2 could have naturally incorporated HIV-1 RNA into its genome when co-infecting a host.
the HIV-1 protein sequence is the same but not the DNA sequence. The moderna DNA sequence is identical. The odds of HIV-1 mixing with bat coronavirus, and then somehow finding the moderna DNA sequence, while not astronomically low, are pretty low.
Well sure, but given that viruses mutate on the order of millions of times a day (since they replicate on the order of gazillions of times a day) it's not terribly unreasonable that this sequence could have developed by accident, even through the relatively winding path you described.
Hell, the entire genetic code of every variant of COVID-19 exists somewhere in the digits of Pi, but that doesn't mean that mathematicians created COVID. This reeks of a slightly more advanced form of numerology to me.
While viruses do mutate a lot, if the sequence were that labile, then you would expect there to be a lot of divergence around the furin cleavage site. We don't really see that, so the site is stable. So, is there something special about that DNA sequence (including the synonymous wobble pairs) that make this a random walk gradient descent minimum? Or is the mutation rate lower than you think.
All I'm saying is when people say "this mutation is super rare therefore it MUST have been manmade!", I'm skeptical. Even the most stable DNA sequences have mutations occasionally, because mutations happen for a bunch of totally uncorrelated reasons. It could have been manmade, but it could have also been made by a stray cosmic ray. To say a particular sequence "proves" that the virus is manmade is...sketchy at best.
This article is worse than useless because it places a pseudoscientific veneer around something (to make it sound plausible) while still being almost certainly wrong.
every major university in the us contains physical copies of all the published articles (IE, people who wanted to learn more, but couldn't afford to subscribe to every journal), annd the rest are on arxiv. Nobody is gatekeeping the evidence.
The reality is that average people simply aren't qualified to deal with the subtlety and complexity of ambiguous data, and the adversaries of public health can just make a video that lies and convince millions of people with seductive, but wrong, ideas. It's extremely rare to have a society which is scientifically savvy enough to understand these sorts of things.
The mutation rate for proofreading ssRNA viruses is, affair, about 10^-7 per nt per replication. If we don’t see a particular variant nt it is (from the top of my head) because it didn’t “survive” the drift events or that there is strong positive selection on the sequence.
(Preemptive) Synonymous mutations are able to be selected, as they affect speed of translation and protein folding due to stalled ribosomes.
How are you calculating those odds? And then once you have calculated them, what are you using to determine your threshold for "low" odds? The odds that a bunch of atoms ended up turning into you and me and the rest of humanity are extremely low or extremely high depending on what your criteria for "low" odds are.
Given the number of bats in the world, the number of human hosts, and probably most importantly, the number of individual virions within each infected host (and therefore the number of replication cycles) is...astronomically high. There are somewhere between 1 and 100 BILLION virions of COVID in each infected person. Now imagine a few thousand bats infected...we are already talking about maybe 1 QUADRILLION different virions (1,000 trillion). It only takes one virion to incorporate some very handy and fitness-increasing HIV-1 RNA into its genome and it is off to the races.
the mutation rate simply is not as high as you imagine it to be. If it were, then we would (likely) see more wobbling around the wobble pairs in the cleavage site coding region.
alright, do you want a specific answer? My guess is it's somewhere around one in 3^{5-6}-ish, for the singular founder event that establishes and fixes that sequence as the canonical sequence for the furin cleavage site of COVID-19.
I never mentioned mutation rate so your gotcha is not as crafty as you might think. Also a mutation in the sense you seem to be implying (random base pair changes due to lack of proof reading) is not really what I am talking about. I am talking about HIV-1 genome being incorporated into SARS-CoV-2 by the host cell or either virus during replication. That is not that crazy when you have quadrillion of replication cycles and a selective pressure to mutate and incorporate new RNA.
we're talking about the odds of the founder event. My guess is odds are about 1 in 3^{5-6} (3-ish wobble options, five to six wobble sites, haven't looked at the sequence to confirm this. Probably someone can do a better analysis based on codon usage in humans. I suggest doing it as an exercise in understanding molecular biology.
I truly have no idea what the answer is, but as someone well versed in economics, econometrics,and statistics, to me the relevant odds are not about independent coin tosses etc. I would like to know P(A|B) where:
A: sequence of 30 nucleotides appears in a virus identified in nature
B: sequence appears in a patent application that predates discovery in nature
Seems to me that would involve a whole bunch of arithmetic, but that ought to be calculable using this database.
The evidence, and more crucially, its meaning and significance are not "verifiable" by 99.999% of the population. That's why things like credentials, resumes, publication history, and peer review exist. That's why we care if someone has promulgated 50 hoaxes in the past, for instance.
So yeah. Credentials do matter, and it matters that this person is not willing to say this out loud, in public, without using a pseudonym.
Alright. I can verify the meaning and significance. I have a PhD in chemical biology, and have done (non-pathogenic) gain of function research. This is the sort of result that I would expect someone to have found if I were given the charge to insert a furin cleavage site into a virus. Proof: https://bmcbiochem.biomedcentral.com/articles/10.1186/1471-2... ,in which I stack-overflow-copy-paste ideas for beneficial mutations from sequences in distant species.
> So yeah. Credentials do matter, and it matters that this person is not willing to say this out loud, in public, without using a pseudonym.
The problem is that we've come so far that him not saying this out loud is not necessarily hinting to him being a charlatan. Plausibly, the cost/benefit analysis by that individum might not make it worth speaking the truth given how heated up the matter is rn. If this sentiment propagates to other topics I fear that larger (and important) parts of the world might drift towards soviet era incompetence
No, you are wrong and you don't know how peer reviewing works.
Reputable journals and other publication outlets use something called "Double Blind peer review" precisely to prevent that the reputation of a researcher could skew the peer review process.
If you want to review and cast a judgement for the points presented in the article, you should do it only by refuting or confirming the content of the text itself. Not because it was written by Einstein or Donald Trump.
> Reputable journals and other publication outlets use something called "Double Blind peer review" precisely to prevent that the reputation of a researcher could skew the peer review process.
Double blind peer review is INCREDIBLY UNCOMMON among high impact biomedical journals
Second, you're missing the key difference, here: peer review is for review and evaluation within the community of qualified scientists.
What I'm talking about is something else: the ability of the reasonably well-informed public to evaluate claims, even though they lack domain expertise.
Some journals and conferences use double blinding. Not all do. In some, the reviewers can see the names of the authors. In others, the authors can propose reviewers, or ask the someone is excluded from review. Journals and conferences may change their rules from one year to the next.
Anyway the blinding, when it exists, is as much to protect the reviewers from retribution by asshole authors, as it is to avoid biasing the reviewers by the reputation of the authors.
I also think the article is wrong, but I disagree with both your points (as scientist myself). Verifiable evidence does not need to be verifiable by some threshold percentage of the population. If that was the case, most math PhD thesis would not be verifiable. Also, writing under a pseudonym is quite understandable given the polemic nature of the topic. If anything, I think we should have more people writing under pseudonyms than less. We could end up with a lot more interesting ideas circulating. The peer review itself must of course be done by trustworthy third parties, but the source of the text need not be.
The credentials don't matter to experts in the field, but this forum is filled with laypeople. No layperson should accept such an article without expert vetting.
No. People can evaluate the evidence for themselves. We do not need authority figures to tell us what to think. This is particularly true when it comes to assessing probabilities, an activity at which most experts have proven hilariously incompetent.
(Musing: What is the probability that a genetics expert's opinion on the probabilities at play would make the discussion less well-informed rather than more well-informed?)
People are so good at evaluating evidence for themselves that we can't end the pandemic due to antivaxers, and there are millions of people who believe the words of Q.
Because they clearly are not qualified in the field based on the quality of the analysis, yet claim as if they have to hide even as a "PhD". Their blog post reads like someone who works on the backend for BLAST but knows nothing about genomes or proteins.
If you were a peer reviewer for a reputable journal (double blind peer reviewing process), you would not need neither the name nor the credentials of the person submitting an article. Actually, scientists and editors would laugh at you for asking for that.
This guy appears to also believe that the omicron variant is not a natural mutation but a lab-made version as well.
BLAST does not contain all genetic material from all viruses or organisms in all of the natural world. There are countless (really, impossible to count) viruses of all kinds out there dancing around in the bodies of all kinds of creatures. There are literally hundreds of trillions of viruses in your body right now (not different kinds but individual viruses). Each of those hundreds of trillions of replications was an opportunity to mutate. Now multiply that by all of the individual animals human or otherwise who could harbor coronaviruses and you have many hundred billions of trillions of chances to generate the gene sequences in question. If I could get billions of trillions of lottery tickets, I'd win the lottery, and here it looks like SARS-CoV-2 won the "does a gene sequence in the virus exist in a patent by moderna" lottery.
if the mutation rate is that high, then why don't we see a ton of sequence variants around the wobble pairs of the furin cleavage site? (Or for any parts of the coronavirus, for that matter?)
no way to answer such a question without real experimental research, and not even sure what you mean by "see", we dont "see" most virus mutations because most mutations don't survive... and certainly aren't sequenced.
And the point isn't that a virus mutates on every replication, but that every replication is a chance for a mutation.
The point is everything here hinges on an argument from probability.
There are more viruses on earth than there are stars in the universe, the astronomical figures are meet with each other. We're not talking different orders of magnitude between the opportunities for a certain mutation and the probability of a certain mutation occuring..
edit: Further, it looks like the sequence exists in some bacteria, meaning it isn't even a question of a random mutation but possibily the incorporation of the sequence from an existing source (which the author claims doesn't exist, but it does, in mycobacterium smegmatis...). this destroys the entire premise of the claim.
Early on one news site I followed kept a long list of all the vaccine candidates that were under development, and there were impressively many. Even the Kazakh Biosafety Research Institute had one.
That kind of worried me. Because even if you think sars-cov-2 is completely natural (and I have no problem accepting that position), have you thought about how many groups around would still have the capacity to make such a virus, if they wanted to?
Even just the fact that "some people believe the virus was made in a lab" can be dangerous. Because what if, say, Kazakhstan's dictator-emeritus Nursultan Nazarbayev believed it? How far-fetched is it that he, or someone like him even in the more nominally democratic parts of the world, would decide, "they fired the first shot, we can't allow there to be a bioweapon gap!" and commandeer the institute to do "dual use" research?
It would be easy to excuse too. They could say, "we're just designing these powerful virus variants to have a vaccine ready in case someone else comes up with the same powerful virus variant", and it would be a reasonable argument, at least as arms race escalation arguments go.
> This guy appears to also believe that the omicron variant is not a natural mutation but a lab-made version as well.
Dismissing everything as far I'd problematic. If you can't prove it and several scientists were baffled by the number of changes and we still donate an answer, you can't criticize someone from believing A while you believe B.
Using the number of BLAST hits as the basis of an argument is about as reliable as using the number of results when searching for a string on GitHub. Without further analysis of the specific sequence and its biological context it can be highly misleading. See a previous Twitter thread on some other purported HIV inserts: https://twitter.com/trvrb/status/1223666856923291648
There also seems to be some circular reasoning the argument. Apparently we can ignore RaTG13 because it’s obviously synthetic, which makes SARS-CoV-2 look even more synthetic. It would be interesting to compare to the BANAL family of SARS-CoV-2-related viruses that are even more closely related to SARS-CoV-2 than RaTG13 [1].
I’m not sure why only viral genomes were searched for the furin cleavage site sequence. Viruses famously exchange genetic material with their host organisms. The “smoking gun” sequence also appears Mycobacterium smegmatis, for example [2].
The GitHub example actually is pretty spot on. If you wrote a non trivial piece of code in 2016, then in 2020 it was used verbatim in another program, what would your conclusion be?
For a non-trivial piece of code, the conclusion might be that it was copied (horizontal code transfer??). But the 6 amino acids is not a "non-trivial piece of code" -- it is a trivial short string of letters. So the more relevant question is, how often do we see "dogcat" in GitHub (468 repository names). Or perhaps something more nonsensical: "ifdlog" (6 letters from "goldfish" reversed): 1 repository and 25 code results. "hsifdl" - 1 code result.
Sure I will not argue that the code showing up somewhere is statistically interesting. What’s interesting is that it showed up exactly once and in a Moderna patent. What’s the probability of that? That of the random strings in the Moderna patent that it would be an exact match to a string in the COVID genome but nowhere else?
I ask a question about dna sequences for SWE interviews. It ends up providing some useful signal. For example questions I ask include: if you have a symbol table with 4 symbols, how many bits are required to uniquely identify each symbol (answer: 2 bits). OK, given this ASCII string 'CTCCTCGG', can you compress it a few bits? (answer: without entropy, run length, or backrefs, you can easily compress DNA encoded in 8-bit ASCII from 8 bits/byte to 2 bits/byte). OK, so, how many bits does an 8-long DNA string require encode (answer: 16 bits). Huh, that's a short.
The rest of the interview is how to move form using a hash table for counting k-mer frequencies to a vector using a minimal perfect hash, and compute k-mer frequencies quickly using a rolling hash. It's a great question because it always throws the CS people off to hear their questions asked with biological structures involved.
some of the answers I gave are the only correct answer.
if somebody was going to go straight to proposing a probabilistic data structure like a bloom count and could code up a simple example in 20-30 minutes, that is also OK (and would lead almost certainly to an immediate hire).
There's a whole branch of CS, bioinformatics, which would be more confused if you did NOT ask the questions with biological structures involved.
The field of string algorithms, or "stringology" as they whimsically call it, is also heavily focused on DNA.
I was planning to study bioinformatics in my youth. Life events intervened and I had to do a regular software engineering degree instead, but I still have an unreasonable love for the Burrows-Wheeler transform.
What you say doesn't really make sense. For example, BWT has nothing specific to do with bioinformatics. It's used in bzip compression, of course, but a wide range of other things, predating its use for genome analysis.
I think the way I'd put it is, DNA sequence analysis benefitted greatly from the wide range of algorithms creatd for string processing- everything from fast string search (boyer moore) to dynamic programming to forward-backward alogirhtm for HMMs. But there's very little, if at all, stringology done that's purely DNA specific.
Yes, I know BWT has nothing specific to do with bioinformatics, but BWT-based compressed indices for bioinformatics is its main practical application today (bzip2 isn't competitive any longer, no one is more sorry about that than me!)
But it's a good example of the thing I say: Even for pure math papers related to BWT, odds are good they use DNA as an example. I wouldn't be "thrown for a loop" at all from your clever bio-themed interview questions, and there are a lot of people like me. Maybe not who love BWT as much, but who in exchange have actually studied bioinformatics.
I'm not sure we're better candidates for it, so you're bringing in some questionable bias.
But since you're a BWT nerd, you'll enjoy my story. I used to work at Google and was well-placed at a key time, sitting near Jeff Dean and Sanjay Ghemawat while launching Google Cloud Genomics. One day, I invited Richard Durbin (one of the innovators in using BW for genomics) to stop by and give Jeff and Sanjay a quick update on his work in sequence compression and indexing. it was great to be a fly on that wall.
At the end of the talk I mentioned that Burrows of Burrows-Wheeler was just down the hall if Richard wanted to chat with him as well. Unfortunately Mike wasn't in the office that day, otherwise I would have been able to cross "Be present when the inventor of an important and subtle algorithm meets a user who has applied it for state of the art research in an unexpectedly productive way" off my bucket list.
As for the bias, sure, but it's fairly small, and I can adjust on the fly to give every candidate a reasonable attempt to solve the problem. I can also give the problem without any reference to biology
for those with SciAm subs, this June 2020 print article is a profile of the first scientist to encounter Covid. She had spent a decade in caves around China investigating SARS and was therefore called on when early reports of Covid arose.
I read this before all the lab leak theories came out, and I'm glad I did because it makes them all seem ridiculous.
She was "called on" to Wuhan WIV where virus first emerged and where she has been working for at least 7 years and where all of her research was funded by EcoHealth Alliance grants? same EcoHealth Alliance which submitted this gain of function grand proposal in 2018 https://theintercept.com/2021/09/23/coronavirus-research-gra... ?
"Also, Dr. Shi was trained by Ralph S. Baric of the University of North Carolina in building “chimera” viruses — taking, for example, the spike protein from a new virus and splicing it to the backbone of a known one like SARS. He invented “no-see-um” techniques that left no trace of the splice."
> the spike protein from a new virus and splicing it to the backbone of a known one like SARS
The spike protein was already a defining feature of SARS-CoV when it emerged and became "known" [1]. That's why it was classified as a coronavirus! [2]
The statement was about grafting a new spike in place of an old one. Where is it implied that SARS doesnt have a spike?
I dont even think that is the important takeaway. The no-see-um editing technique can apply to more than just spike proteins. I dont think its being implied that thats what happened.
That covid was made in a lab? Or did covid and moderna independently reach the same "conclusion" through parallel evolutionary pressures (in nature and in research)?
The second possibility is the main objection I have to this post. The lottery arguments may actually be good enough to rule out chance, but ruling out chance isn't the same as ruling out other explanations which can make things extremely non-random.
As other commenters point out, this "analysis" is a joke. An E()-value of 282 says we would expect a match of this short sequence against a completely random library of the same size 282 times. Even with BLASTp's parameter correction for short sequences, NO short sequence will generate a statistically significant score.
Let's try a more sensible approach. Let's see how often the sequence occurs in other databases. The author's 1/20^6 (1/64,000,000) probability calculation omits the fact that every database search considers many millions of possible alignments (a crude calculation would be 6 * the database length, or 6 * 1.6E7/20^6 ~ 10000E6/64E6 = 156 times by chance), and we're shown that in a 5,000,000 protein/1.6E7 residue database, 100% matches are seen several dozens of times, as expected.
How about other protein sets? Search the human protein set (19,000 proteins), there are no 6 amino acid 100% identical matches. The best matches are 100% of 5 residues, or 83% identity over 6 (with an E()-value of 22).
How many proteins do we need to search to find an exact full-length match? (we have dozens of matches in a 5,000,000 viral protein database. and none in 19,000). If you search all the NCBI landmark sequences (~500,000, non-redundant), there are 3 identical matches (E()-value: ~44) in Drosophila, Corn, and Strep. pneumoniae. So this sequence is not very special -- as the E()-value statistics indicate, it happens all the time by chance. (If it happens 3 times in a 500,000 sequence non-redundant database by chance, we would expect it to happen 30-100's of times in a 5,000,000 redundant viral database by chance, which is what we see.)
Six amino acid non-significant matches are evidence that properly done statistics are accurate, and pretty much nothing else.
It's not only the numbers. 1 out of 282 with the one being Moderna is much more significant. The null hypothesis is important. It's not just the odds reporting. Anybody can pass by your door accidentally
Less likely it is someone that threatened you last night. Same odds, different situations.
It is not clear to me why it is much more significant (or how much more). There is confirmation bias here. There would be no "report" if Moderna had not been associated with the sequence. So how many other, perhaps longer, matches were there to other patented sequences with no connection to Moderna? If there were hundreds (which there probably were), it is much less clear what the significance of one of them being in a Moderna patent is, particularly if Moderna was trying to make an HIV vaccine, and the sequence is in HIV (which it is).
In my cursory encounters with numerology, I have learned that if you try enough hypotheses, something will always pop out. Could this be the case here too, or is the gun-smoke so thick that no cherry picking can explain it?
Why is the native hiv sequence (which has had a lot of convergence potential since hiv mutates, a lot) different, and why did it convergently evolve instead to the nucleotide sequence of moderna's artificial furin site (designed for hiv), which to be fair, is subjected to industrial selection, which would be (I would guess, but don't know) weaker than the natural one due to selection sampling limitations?
> *As of writing this those links are still up which at 12 months is pretty good going for any article that dares challenge the drivel propagandised by our beloved “free press [sponsored by pharma]”.
Well this is a rough start. “This information is being suppressed, which you can tell by the fact that these linked articles are still available.” The utter lack of critical thinking doesn’t make me expect much from the rest of this.
1. Why do we care so much about a furin cleavage site? The author makes a big deal out of this and I'm not sure why. It seems like he specifically looked at the gene sequence at a furin cleavage site and then said: hey look, this occurs at a furin cleavage site! It must mean something! (Thanks to etaoins for finding at least some evidence that furin cleavage sites are not, in fact, unusual in Coronaviruses, although I have not read the paper: https://www.sciencedirect.com/science/article/pii/S187350612...)
2. What does HIV-1 have to do with anything? That is, if these sequences all matched virus XYZ instead, would that also be some sort of smoking gun? The top 3 sequences all seem to be very common among lots of viruses (rotavirus, measles, coronaviruses, and HIV all show up in the top 3 lists). So he seems to have found a single sequence that lots and lots of viruses have (at least variants of), and that one sequence is the first 3 rows of his table. Am I missing something?
3. If we ignore every result entered after Feb 2020, aren't we throwing out a lot of sequences that were entered specifically because people were suddenly all very interested in Coronaviruses? If he's making claims like "these don't appear in any other Coronavirus", the "except those we found after everyone started looking at Coronaviruses much more" somewhat detracts from that, doesn't it?
It feels a little bit like finding a passage "He froze, trembling with fear" and saying: the English language has 1 million words, so the probability of finding these 5 words in order is (1 million)^5. Therefore if we find it in two novels, one must have plagiarized the other. But not really though, because "he" and "with" are extremely common words, and "froze", "trembling", and "fear" are very highly correlated to each other. I think it's pretty clear his first 3 examples are all highly correlated.
His last example (and the title of this post) seem a little harder to explain to me as a layperson, but I'd note that for a 30-nucleotide sequence, there are 3^30 other sequences that are only a single mutation away. We again see Rotavirus on his list. How far away is that Rotavirus sequence, and do we think it prohibitively unlikely this sequence came from Rotavirus?
Chemical biology Phd, did work on (non-pathogenic) gain of function research here.
1. We care about a furin cleavage site, because it sticks out like a sore thumb. It doesn't exist in the closest family of coronaviruses to SARS-COV-19, yet exists in distant relatives. This is the exact sort of thing you fuck around with if you are doing GOF (pathogenic or otherwise) research -- you know it's likely to be tolerated by the parent scaffold because relatives feature it and aren't completely unstable -- and you see the vague outlines of a correlation with function.
Furin cleavage sites are also something a bioengineers like to engineer because we are familiar with them (we often create furin cleavage sites for protein production -- to liberate the purified protein from whatever scaffolding we use to build the protein -- or for analytical purposes, because you can get good evidence that you built what you built if it gets chopped in two).
Finally, we know that a nonprofit associated with the Wuhan Institute of Virology proposed specifically to engineer furin cleavage sites into coronaviruses (in association with WIV, but with the research done in the US), which proposal was rejected by DARPA.
2. HIV doesn't have anything to do with anything except that this is what the protein sequence of the spike matches up with. This is ~6 amino acids, which is admittedly not a whole lot, but not all furin cleavage sites look alike. If I am reading this blogpost correctly, the HIV-1 DNA sequence (which can represent the same protein sequence with worst case ~60% divergence) doesn't match up with the coronavirus, but the moderna patent sequence does (and it's exact). The moderna patent furin cleavage sequence known to be based on the HIV-1 sequence.
3. You ignore those results because NCBI will send you a hell of a lot of COV-19 sequences, because guess what, we've been sequencing the shit out of COV-19. But your criticism of the "statistics" that come out is 100% correct. It's possible that in the last two years have found a "missing link" that will be filtered out by those criteria, though one suspects that if such a thing were found, it would have made some pretty big headlines.
So, if we do a dispassionate analysis without considering faulty application statistics to the pre-2020-only sequence counts, and simply sticking to the existence or inexistence of sequences as per the database we are left with a few possibilities:
1.
COV-19 precursor protein intermixed with HIV-1 and did an uptake of the DNA (this is totally possible but rather rare, even given that HIV-1 is rampant in the wuhan-ish part of the world), then somehow converged on the moderna sequence (let's say 1 in 250 odds).
2.
Some (prior-to-2020 unknown or unregistered) DNA sequence that happens to exactly match up the moderna sequence, by chance, got merged with some wild Coronavirus sequence to yield covid-19.
3.
- someone came up with the idea of fusing furin into coronaviruses (true if you believe the leaked DARPA grant application, btw I have read the whole thing and the level of detail and its similitude to grants I have read and written would have been really hard, but not impossible, to fake, especially by a state actor.)
- someone went ahead and did the research on a slightly different coronavirus than proposed in the DARPA leak even without grant money (this does happen in biology, I'm probably on the tail end of the distribution but out of 10 years in science only 1 was working on previously proposed research, the other 9 were on research I was gathering preliminaries for to write grants)
- the postdoc or grad student assigned to this went and collected a bunch of furin cleavage sites, probably off of NCBI blast, and stack-overflow-copy-pasted them into the coronavirus, of which one was the site from the moderna patent.
- the moderna sequence randomly happened to be the "winner" and spread around.
I've noticed that you are responding to many people that blatantly ridicule the original article and the author for writing behind a pseudonym. They are just trying to manipulate uninformed people towards ignoring this information.
The key to all this is that Coronaviruses have been studied for decades and there is no single coronavirus with that furin cleavage and HIV matches on the bindings.
After testing thousands of bats, no one has found an animal with COV-19 and, even worse, it is not easy to make bats to get infected by COV-19.
The fact that this short strings are in patents by one of the main manufacturers of the vaccines and the other ones are part of HIV which are directly connected to the Early Life of Fauci might be coincidental, but there is nothing that indicates that this is a natural virus.
Hold on. Let's be precise. Other coronaviruses do have furin cleavage sites. No other coronaviruses have this particular furin cleavage site.
> but there is nothing that indicates that this is a natural virus.
Let's also be clear here. You don't have to have anything that indicates that this is a natural virus; there is no reason why (as some of the other rabid commenters on the "other" side keep saying) this couldn't be natural. It would take some (very possible) stretches to get there. Biology does stretch sometime, there is no occam's razor, consider the platypus.
In the end I think one has to look at the proposed chain of events. My belief and let me be clear that this is a belief is that it's a lab leak. From when this started, I have half-jokingly said that this is how I would have mishandled the virus (I'm pretty clutzy in the lab - imploded no less than two dewars), and over time evidence has trickled in making the lab leak hypothesis stronger [0], while the observations you would expect have come out that makes the 'natural' hypothesis stronger have not surfaced, so in my mind, the trendline has been towards lab leak. Over time, the 'missing chain of events' that would have to have happened for it to be a lab leak are closing in. But sure, we could find a credible missing link for the natural hypothesis tomorrow and I would have to revise my beliefs. Honestly, the way the CPC has locked down information about what happened, we will almost certainly never know for sure and either way it will likely have to remain a belief.
[0] e.g. the leaked DARPA grant, information that the French refused to certify WIV lab, genetics on lao and yunnan viruses, similar lab leak happening in Taiwan, etc.
I forgot to answer something, rotavirus "appears on the list" is only there because it's a synthetic SARS virus-component expression system that's sitting on a commonly used rotavirus-based expression cassette.
A large part of this blog post's ridiculous assertion lies either on ignorance or on the false assumption that horizontal gene transfer [0] does not exist in Nature:
> It’s easy enough to change a single nucleotide (a single point mutation or SNP) or even insert or delete nucleotides (less common) but to insert 20 or 30 nucleotides with a code that works? Nope, that has to come from another virus or else it’s been done in a lab.
The original preprint which noted similarity of the HIV-1 sequences and those in the COVID spike protein, and on which this blog author's claims rest, was voluntarily withdrawn by its authors. [1]
Moreover, the assertions of uniqueness have been thoroughly addressed in the published article, "HIV-1 did not contribute to the 2019-nCoV genome" [2] :
> For any virus to obtain additional insert sequences from other organisms, it requires that it has direct interactions with other organisms, most likely through homologous or non-homologous recombination. [...] On the contrary, these motifs are widely present in various mammalian cells and so it will be more likely for bat CoV viruses to gain those motifs from the genomes of their infected cells if recombination indeed occurs.
That rebuttal article clearly suggests several feasible sources of the gene in Nature. That surely is more plausible than suggesting that Moderna or some other group nefariously created the COVID genome _in vitro_ before unleashing it upon humanity.
I deeply enjoy Hacker News for its ability to surface different ideas and sources of information that I would otherwise never find, but this article is simply ridiculous.
The really ridiculous thing is that we're expected to "explain it like I'm 5" to smart computer hackers who don't know biology and constantly try to poke holes in mainstream science without taking the time to understand the complexity and subtlety in that field.
I'm not really sure who you're talking about here. Who is expecting you to explain things to them simply? Informative comments are always appreciated, but I'm not sure why you're taking this as some personal obligation. Who is poking holes in mainstream science? I don't really see that happening in these comments.
I follow a lot of threads on the site, especially ones that appear to be COVID disinformation. And I see a lot of hackers asking "well, I read this <disinformation> and it made sense to me. Can you explain why it's not true?"
This is pretty common. The reason I take it as obligation is that if there aren't people fighting disinformationn, the disinformation will win, because it tends to be full of simple, seductive, but. wrong ideas that naive people fall for.
Well, it's possible the author believes what they wrote in which case it would be "false information believed to be true intended to teach" but I'm pretty sure this is active misinformation.
I'm not sure which particular thing public health people said that you're complaining about in the last part of the section- but I don't really think that public health leaders who communicate things which aren't 100% correct are dissemination disinformation. That's sort of a pedantic and uninteresting argument to have about the philosophical nature of information.
As a computer science guy I did my own ELI5, found out how insertion sequences work, found the paper about HIV-1 and 2019-nCov, but it is true, it takes time to understand.
Even without spending time my heuristics is that I find biases in the claim and find out how referenced it is. This one had plenty red flags, including dismissing the RaTG13 results claiming that they were just put into the database after people started questioning, the adversarial tone, presumptions and so on. Conspiracy theorists use recognizable heuristics, but their motivation is to seek attention, for which there is large competition. They want to be relevant, but who doesn't. Conspiracy theorists are not really into publishing a paper either, a blog post, video or tweet is all you get, it's not about the truth and it is easier to avoid being found out this way, though social media these days have fact checkers, so they have to fly under the radar or keep to the social circles that are not bothered by fact checkers.
Knowing about biases and how to recognize them is a vital skill, should be taught in schools. Everyone needs a working bullshit meter.
This appears to be a disease among technical people. The number of people who work in something like software and have decided that they can effectively evaluate research in a completely different field is frightening.
I get that there's a lot of fear around the idea that Covid-19 was engineered, and the potential fallout that could occur if it was proven to be fact. But I expect that almost all governments are now working under the assumption that it was, because they would be extremely foolish not to. So this is all a moot point -- at the top level, the working theory must be the lab-leak theory.
And so, fast vaccine production must be a priority. The development of drugs to treat viral infections and their manufacture must also be a priority. The ability to detect carriers and quickly shut down travel must also be a priority. And how do we enforce quarantine in populations that are resistant to it? All of these concerns are, I am certain, at the forefront in the mind of any Western government to be certain, but likely any government anywhere.
Because regardless of IF Covid-19 was created in a lab, it COULD HAVE BEEN created in a lab. And we all got caught with our pants down.
Funnily enough those screaming it was engineered loudest also were the ones saying it’s just the flu and we should react to it by not wearing masks.
I propose that what we really see is that humans have a very hard time without someone to blame (hence victim blaming, hence religion and a whole bunch of other coping mechanisms).
It doesn't look like that nucleotide sequence is actually unique to both SARS-CoV-2 and Moderna's 2015/2016 patents. Just BLASTed that sequence with the "gss" database and got hits¹ for a bunch of different species of grass (plus some other hits for everything from bacteria to a species of dolphin). Tried again with the "RefSeq_Gene" database (ostensibly the human genome, AFAICT) and got a bunch of hits² for that, too.
I ain't a geneticist, so it goes without saying that I don't really know what I'm doing and I'm probably using BLAST wrong, but it seems to me like (EDIT: something very close to) CTCCTCGGCGGGCACGTAG is pretty dang common even within the human genome (let alone other species), and it doesn't seem far-fetched to me that SARS-CoV-2 might've yanked (EDIT: something very close to) that sequence from one of its hosts at some point.
EDIT: I forgot to check the completeness of the matches, and it does look like the hits tend to have one or two nucleotides that don't match. Still, it's close enough that the article doesn't seem like it presents much of a smoking gun.
Yep, you're right (and edited above to reflect that).
After accounting for that, and after realizing that BLAST only returns the first 100 results, I searched on the "nt" database (where the article found "only" SARS-CoV-2 results) while excluding SARS-CoV-2 (taxid:2697049), "synthetic construct" (taxid:32630), and "other sequences" (taxid:28384). This¹ produced not just a lot of hits, but a lot of exact (i.e. 19/19) hits (mostly various microbes) - further suggesting that even the exact sequence in question ain't unique.
I also ran a search on "nt" strictly on SARS-CoV (i.e. the original SARS virus)², and while no single result completely matches the query, there are a lot of results that could be concatenated/combined to produce the query - and while I'm (again) no geneticist, it doesn't seem to me like some sort of impossibility for that concatenation to happen through mutations in the wild.
Unless there is a transposable element between them, Just so concatenations like that are two independent splicing events so you roughly expect a quadratic likelihood reduction (square the rarity)
Which would still be a lot more likely than "random chance" (as the article puts it - i.e. each of those 19 nucleotides being randomly selected), right?
Not really, the opposite would be true—with the exact same evidence—if it was the government's position that the virus was made in a lab.
Then you'd have Substack's talking about how the virus could have been naturally occurring, that there's no reason it had to be made in a lab, that it's just the government blaming China, etc.
I wish there was a word for beliefs that supposedly make use of evidence, but can easily use the exact same evidence to argue for the polar opposite belief.
>What this means is that there is no virus known to man that has this particular sequence in its genome prior to the discovery of SARS-Cov-2. So where on earth has it come from?
169 comments
[ 0.24 ms ] story [ 237 ms ] thread>Posted by 'version_five'
"This sounds like one of those "bible code" things. As someone who knows nothing about DNA sequences, what is the probability that two codes could be matched? (In a birthday paradox sense where you can go looking anywhere)"
>Posted by 'whoomp12342'
"isn't it well known that one of the main reasons why we got a vaccine so quickly was because we had done a ton of research when SARS originally broke out?"
An unrelated threatening comment would have gotten the unrelated comment killed, but not the main posting.
If the main posting was flagged, that is because some number of users with enough karma to flag clicked the "flag" link on the post.
The genome in question is about 30000 bases long, which isn't nearly long enough for the outcome space size issue to come into play. Assuming independence, the probability is still very remote, effectively zero.
However, there is in no way any expectation of independence. SARS-CoV-2 is obviously closely related to SARS-CoV-1, which is obviously closely related to 2015/2016 ongoing vaccine research. (And perhaps gain of function research.)
>this sequence
>this
>sequence
Furin clevage sites haven’t been found elsewhere in SARS-Cov-2’s Sarbecovirus subgenus.
And by "goalposts" I mean that if you recall, the initial allegation was that COVID was a bioweapon developed by the Chinese. Then it was negligently leaked from a lab doing gain of function research funded by Obama. Then it was just leaked. People seem to want to believe this theory regardless of when facts get in the way. A year ago, there wasn't much evidence against it. Now there are animal finds in Laos that make some of the connection. Not proof, but positive evidence. From the field.[1] Not from cherry-picked database searches.
At some point, you have to ask yourself with Ockham's razor: if it really is this simple, then why isn't it widely accepted? Has China gotten to everyone? If so, then why can I post this to my substack? Sometimes the expert opinion is right. I guess sheeple like me will never know. Except in reality, I'm quite open to this hypothesis, I am already a China hawk so I need no motivation there. I just don't think there's as much evidence as people who want to believe this think there is.
For what it's worth, using the model of radius->radii for virus->*virii makes sense until you realize radius has an extra 'i' already. Worse, it's a second declension neuter despite the -us and so should be vira just like modicum->modica, whereas radius is second declension masculine.
[1] https://www.nature.com/articles/d41586-021-02596-2
https://www.evidenceinvestor.com/the-twenty-dollar-bill/
Nevermind the diplomatic implications. If, hypothetically, tomorrow the official stance of the U.S. Government changed to hold that China was responsible for the manufacture of the virus, that despite the part the U.S. had in funding it being facilitated by failures in process at NIH, this would signal a sizable shift in attitude toward the Chinese, and could very well escalate tensions to a degree no one has a stomach for.
>There's no "establishment" conspiracy to suppress evidence of the lab leak theory. The Biden administration ordered further investigation as one of their first acts. No suppression there.
You have no idea how much classified data and information may exist. There is a whole corpus of knowledge kept out of the public's view for years because of how the Executive feels about how much more complicated it would make their lives. I would not dismiss official perception management as quickly as you do. The government does not hint. It does not offer, and leaves it to the public to do their own research until such time as the public largely figures it out independently; and then still will not confirm or deny. Look into how the body of knowledge around nuclear weapons was cobbled together by uncleared or non-government affiliated researchers. FOGBANK was such a closely guarded secret, it had to have it's manufacturing process reserved. Same with various B-1 part manufacturing processes.
I have no idea how old you are, or how much time you spend observing how groups of people work and behave emotionally, but it's fairly obvious to me that this type of finding is something no one with an appreciation for second order effects wants to let out of the bottle without a lot of thought, or blind faith it is the RIGHT thing to do; least of all those virologists who either truly believe GoF research is necessary, or those that wish to conduct research without a billion hoops to jump through. Further, not all the old-timers from WWII are dead yet, and those with parents who lived through the Cold War are fairly certain that if it comes down to a WWIII, there may not be a return to any semblance of normality in the aftermath. This is exactly the kind of flashpoint in history that resonates with lighting that kind of fuse.
If you believe in all the conspiracy theories, you're a sucker, but if you believe there are ZERO conspiracy theories, you're also a sucker.
https://news.ycombinator.com/newsguidelines.html
You (by which I mean not you personally, but everyone here) need to follow the rules regardless of what other people are doing. Taking the intended spirit of the site more to heart would be helpful: https://news.ycombinator.com/newsguidelines.html
This would be a particularly useful discovery, because it would dramatically limit the candidates for who did the gain of function research to organizations licensed this Moderna lineage.
- Humans and their pathogens
- Economically important organisms (e.g. crops and farm animals) and their pathogens
- Model organisms
For example, compare the number of Genebank sequences for HIV-1 [1] versus Feline immunodeficiency virus [2]. HIV-1 is additionally problematic because it has an extremely high mutation rate [3], making it more likely for a random sequence to match some sequenced HIV-1 genome.
[1]: https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?mod...
[2]: https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?mod...
[3]: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4574155/
So unless you go to a lot of trouble (e.g. the NCBI landmark database), most sequence databases are a mixture of 10's to 10,000's of very closely related sequences (not a random sample), from a non-random sample of organisms (we tend to sequence pathogens).
Protein and DNA sequence databases are very large, but they are not random.
Edit: to elaborate, certain amino acid sequences are seen again and again, and others are practically impossible to see in nature because amino-acids fold into 3D proteins...and each amino acid is like a Lego piece. There is a lot more detail needed to understand fully how tertiary and 4try protein structures form...but it is also possible to understand it decently well after an undergraduate degree in biology, chemistry, biochemistry, etc. A PhD in the field knows this inside and out. If I had to guess, the author of this post has a PhD in something like statistics or computer science, and thinks that they can apply high school level math to two fields (molecular biology and biochemistry) that they do not even have a high school level of understanding of.
To make a very stretched analogy (I am a doctoral student in the life sciences and a hobby programmer), this blog post is like saying that some Java source code is stolen because four of the class names are the same between two projects, and then using the number of possible characters in the class names and the length of the name to run some statistics. The problem being that no one has class names like "AeNOQ92bA"...in fact the vast majority of 8 character sequences will never be class names. Just like the vast majority of amino acid sequences (likely) do not exist in nature. And then you dig deeper and find out the class name is something like "MainDashboardSupervisorTree" (I do not know Java so forgive me). Then you also find out that the author of both classes...is the same guy who moved companies and likes particular naming convection, but never meant to copy stuff word for word. Similar to how SARS-CoV-2 could have naturally incorporated HIV-1 RNA into its genome when co-infecting a host.
Hell, the entire genetic code of every variant of COVID-19 exists somewhere in the digits of Pi, but that doesn't mean that mathematicians created COVID. This reeks of a slightly more advanced form of numerology to me.
The reality is that average people simply aren't qualified to deal with the subtlety and complexity of ambiguous data, and the adversaries of public health can just make a video that lies and convince millions of people with seductive, but wrong, ideas. It's extremely rare to have a society which is scientifically savvy enough to understand these sorts of things.
(Preemptive) Synonymous mutations are able to be selected, as they affect speed of translation and protein folding due to stalled ribosomes.
Given the number of bats in the world, the number of human hosts, and probably most importantly, the number of individual virions within each infected host (and therefore the number of replication cycles) is...astronomically high. There are somewhere between 1 and 100 BILLION virions of COVID in each infected person. Now imagine a few thousand bats infected...we are already talking about maybe 1 QUADRILLION different virions (1,000 trillion). It only takes one virion to incorporate some very handy and fitness-increasing HIV-1 RNA into its genome and it is off to the races.
I truly have no idea what the answer is, but as someone well versed in economics, econometrics,and statistics, to me the relevant odds are not about independent coin tosses etc. I would like to know P(A|B) where:
A: sequence of 30 nucleotides appears in a virus identified in nature
B: sequence appears in a patent application that predates discovery in nature
Seems to me that would involve a whole bunch of arithmetic, but that ought to be calculable using this database.
So yeah. Credentials do matter, and it matters that this person is not willing to say this out loud, in public, without using a pseudonym.
The problem is that we've come so far that him not saying this out loud is not necessarily hinting to him being a charlatan. Plausibly, the cost/benefit analysis by that individum might not make it worth speaking the truth given how heated up the matter is rn. If this sentiment propagates to other topics I fear that larger (and important) parts of the world might drift towards soviet era incompetence
Reputable journals and other publication outlets use something called "Double Blind peer review" precisely to prevent that the reputation of a researcher could skew the peer review process.
If you want to review and cast a judgement for the points presented in the article, you should do it only by refuting or confirming the content of the text itself. Not because it was written by Einstein or Donald Trump.
Double blind peer review is INCREDIBLY UNCOMMON among high impact biomedical journals
Second, you're missing the key difference, here: peer review is for review and evaluation within the community of qualified scientists.
What I'm talking about is something else: the ability of the reasonably well-informed public to evaluate claims, even though they lack domain expertise.
Anyway the blinding, when it exists, is as much to protect the reviewers from retribution by asshole authors, as it is to avoid biasing the reviewers by the reputation of the authors.
(Musing: What is the probability that a genetics expert's opinion on the probabilities at play would make the discussion less well-informed rather than more well-informed?)
Why, according to you are they worth and not worth examining?
BLAST does not contain all genetic material from all viruses or organisms in all of the natural world. There are countless (really, impossible to count) viruses of all kinds out there dancing around in the bodies of all kinds of creatures. There are literally hundreds of trillions of viruses in your body right now (not different kinds but individual viruses). Each of those hundreds of trillions of replications was an opportunity to mutate. Now multiply that by all of the individual animals human or otherwise who could harbor coronaviruses and you have many hundred billions of trillions of chances to generate the gene sequences in question. If I could get billions of trillions of lottery tickets, I'd win the lottery, and here it looks like SARS-CoV-2 won the "does a gene sequence in the virus exist in a patent by moderna" lottery.
And the point isn't that a virus mutates on every replication, but that every replication is a chance for a mutation.
The point is everything here hinges on an argument from probability.
There are more viruses on earth than there are stars in the universe, the astronomical figures are meet with each other. We're not talking different orders of magnitude between the opportunities for a certain mutation and the probability of a certain mutation occuring..
edit: Further, it looks like the sequence exists in some bacteria, meaning it isn't even a question of a random mutation but possibily the incorporation of the sequence from an existing source (which the author claims doesn't exist, but it does, in mycobacterium smegmatis...). this destroys the entire premise of the claim.
And even if they survive the selection pressure is immense.
That kind of worried me. Because even if you think sars-cov-2 is completely natural (and I have no problem accepting that position), have you thought about how many groups around would still have the capacity to make such a virus, if they wanted to?
Even just the fact that "some people believe the virus was made in a lab" can be dangerous. Because what if, say, Kazakhstan's dictator-emeritus Nursultan Nazarbayev believed it? How far-fetched is it that he, or someone like him even in the more nominally democratic parts of the world, would decide, "they fired the first shot, we can't allow there to be a bioweapon gap!" and commandeer the institute to do "dual use" research?
It would be easy to excuse too. They could say, "we're just designing these powerful virus variants to have a vaccine ready in case someone else comes up with the same powerful virus variant", and it would be a reasonable argument, at least as arms race escalation arguments go.
Dismissing everything as far I'd problematic. If you can't prove it and several scientists were baffled by the number of changes and we still donate an answer, you can't criticize someone from believing A while you believe B.
There also seems to be some circular reasoning the argument. Apparently we can ignore RaTG13 because it’s obviously synthetic, which makes SARS-CoV-2 look even more synthetic. It would be interesting to compare to the BANAL family of SARS-CoV-2-related viruses that are even more closely related to SARS-CoV-2 than RaTG13 [1].
I’m not sure why only viral genomes were searched for the furin cleavage site sequence. Viruses famously exchange genetic material with their host organisms. The “smoking gun” sequence also appears Mycobacterium smegmatis, for example [2].
[1]: https://www.nature.com/articles/d41586-021-02596-2 [2]: https://twitter.com/soychicka/status/1243547603746410500
Random things occur at random.
The rest of the interview is how to move form using a hash table for counting k-mer frequencies to a vector using a minimal perfect hash, and compute k-mer frequencies quickly using a rolling hash. It's a great question because it always throws the CS people off to hear their questions asked with biological structures involved.
if somebody was going to go straight to proposing a probabilistic data structure like a bloom count and could code up a simple example in 20-30 minutes, that is also OK (and would lead almost certainly to an immediate hire).
The field of string algorithms, or "stringology" as they whimsically call it, is also heavily focused on DNA.
I was planning to study bioinformatics in my youth. Life events intervened and I had to do a regular software engineering degree instead, but I still have an unreasonable love for the Burrows-Wheeler transform.
I think the way I'd put it is, DNA sequence analysis benefitted greatly from the wide range of algorithms creatd for string processing- everything from fast string search (boyer moore) to dynamic programming to forward-backward alogirhtm for HMMs. But there's very little, if at all, stringology done that's purely DNA specific.
But it's a good example of the thing I say: Even for pure math papers related to BWT, odds are good they use DNA as an example. I wouldn't be "thrown for a loop" at all from your clever bio-themed interview questions, and there are a lot of people like me. Maybe not who love BWT as much, but who in exchange have actually studied bioinformatics.
I'm not sure we're better candidates for it, so you're bringing in some questionable bias.
But since you're a BWT nerd, you'll enjoy my story. I used to work at Google and was well-placed at a key time, sitting near Jeff Dean and Sanjay Ghemawat while launching Google Cloud Genomics. One day, I invited Richard Durbin (one of the innovators in using BW for genomics) to stop by and give Jeff and Sanjay a quick update on his work in sequence compression and indexing. it was great to be a fly on that wall.
At the end of the talk I mentioned that Burrows of Burrows-Wheeler was just down the hall if Richard wanted to chat with him as well. Unfortunately Mike wasn't in the office that day, otherwise I would have been able to cross "Be present when the inventor of an important and subtle algorithm meets a user who has applied it for state of the art research in an unexpectedly productive way" off my bucket list.
As for the bias, sure, but it's fairly small, and I can adjust on the fly to give every candidate a reasonable attempt to solve the problem. I can also give the problem without any reference to biology
I read this before all the lab leak theories came out, and I'm glad I did because it makes them all seem ridiculous.
https://www.scientificamerican.com/article/how-chinas-bat-wo...
"Also, Dr. Shi was trained by Ralph S. Baric of the University of North Carolina in building “chimera” viruses — taking, for example, the spike protein from a new virus and splicing it to the backbone of a known one like SARS. He invented “no-see-um” techniques that left no trace of the splice."
source https://donaldgmcneiljr1954.medium.com/how-i-learned-to-stop...
https://www.researchgate.net/publication/8119695_Development...
The spike protein was already a defining feature of SARS-CoV when it emerged and became "known" [1]. That's why it was classified as a coronavirus! [2]
Edit: Appears I have misread, see reply
[1] https://www.science.org/doi/abs/10.1126/science.1086418
[2] https://en.wikipedia.org/wiki/Coronavirus#Structure
The statement was about grafting a new spike in place of an old one. Where is it implied that SARS doesnt have a spike?
I dont even think that is the important takeaway. The no-see-um editing technique can apply to more than just spike proteins. I dont think its being implied that thats what happened.
That covid was made in a lab? Or did covid and moderna independently reach the same "conclusion" through parallel evolutionary pressures (in nature and in research)?
Let's try a more sensible approach. Let's see how often the sequence occurs in other databases. The author's 1/20^6 (1/64,000,000) probability calculation omits the fact that every database search considers many millions of possible alignments (a crude calculation would be 6 * the database length, or 6 * 1.6E7/20^6 ~ 10000E6/64E6 = 156 times by chance), and we're shown that in a 5,000,000 protein/1.6E7 residue database, 100% matches are seen several dozens of times, as expected.
How about other protein sets? Search the human protein set (19,000 proteins), there are no 6 amino acid 100% identical matches. The best matches are 100% of 5 residues, or 83% identity over 6 (with an E()-value of 22).
How many proteins do we need to search to find an exact full-length match? (we have dozens of matches in a 5,000,000 viral protein database. and none in 19,000). If you search all the NCBI landmark sequences (~500,000, non-redundant), there are 3 identical matches (E()-value: ~44) in Drosophila, Corn, and Strep. pneumoniae. So this sequence is not very special -- as the E()-value statistics indicate, it happens all the time by chance. (If it happens 3 times in a 500,000 sequence non-redundant database by chance, we would expect it to happen 30-100's of times in a 5,000,000 redundant viral database by chance, which is what we see.)
Six amino acid non-significant matches are evidence that properly done statistics are accurate, and pretty much nothing else.
(eg https://i1.wp.com/www.ghgossip.com/wp-content/uploads/2020/0...)
Nature favors sequences that work. Try a billion variants, keep the one that actually helps. We don't see the 999,999,999 failures.
Well this is a rough start. “This information is being suppressed, which you can tell by the fact that these linked articles are still available.” The utter lack of critical thinking doesn’t make me expect much from the rest of this.
1. Why do we care so much about a furin cleavage site? The author makes a big deal out of this and I'm not sure why. It seems like he specifically looked at the gene sequence at a furin cleavage site and then said: hey look, this occurs at a furin cleavage site! It must mean something! (Thanks to etaoins for finding at least some evidence that furin cleavage sites are not, in fact, unusual in Coronaviruses, although I have not read the paper: https://www.sciencedirect.com/science/article/pii/S187350612...)
2. What does HIV-1 have to do with anything? That is, if these sequences all matched virus XYZ instead, would that also be some sort of smoking gun? The top 3 sequences all seem to be very common among lots of viruses (rotavirus, measles, coronaviruses, and HIV all show up in the top 3 lists). So he seems to have found a single sequence that lots and lots of viruses have (at least variants of), and that one sequence is the first 3 rows of his table. Am I missing something?
3. If we ignore every result entered after Feb 2020, aren't we throwing out a lot of sequences that were entered specifically because people were suddenly all very interested in Coronaviruses? If he's making claims like "these don't appear in any other Coronavirus", the "except those we found after everyone started looking at Coronaviruses much more" somewhat detracts from that, doesn't it?
It feels a little bit like finding a passage "He froze, trembling with fear" and saying: the English language has 1 million words, so the probability of finding these 5 words in order is (1 million)^5. Therefore if we find it in two novels, one must have plagiarized the other. But not really though, because "he" and "with" are extremely common words, and "froze", "trembling", and "fear" are very highly correlated to each other. I think it's pretty clear his first 3 examples are all highly correlated.
His last example (and the title of this post) seem a little harder to explain to me as a layperson, but I'd note that for a 30-nucleotide sequence, there are 3^30 other sequences that are only a single mutation away. We again see Rotavirus on his list. How far away is that Rotavirus sequence, and do we think it prohibitively unlikely this sequence came from Rotavirus?
let's just say some number of scientists love jargon and they love to bandy about "furin cleavage site" and "smoking gun" because it sounds cool.
For your second question, HIV-1 doesn't have anything to do with this.
But then why is 40% of his article and 3/4 of the rows in the table devoted to the similarity between SARS-CoV-2 and HIV-1?
1. We care about a furin cleavage site, because it sticks out like a sore thumb. It doesn't exist in the closest family of coronaviruses to SARS-COV-19, yet exists in distant relatives. This is the exact sort of thing you fuck around with if you are doing GOF (pathogenic or otherwise) research -- you know it's likely to be tolerated by the parent scaffold because relatives feature it and aren't completely unstable -- and you see the vague outlines of a correlation with function.
Furin cleavage sites are also something a bioengineers like to engineer because we are familiar with them (we often create furin cleavage sites for protein production -- to liberate the purified protein from whatever scaffolding we use to build the protein -- or for analytical purposes, because you can get good evidence that you built what you built if it gets chopped in two).
Finally, we know that a nonprofit associated with the Wuhan Institute of Virology proposed specifically to engineer furin cleavage sites into coronaviruses (in association with WIV, but with the research done in the US), which proposal was rejected by DARPA.
2. HIV doesn't have anything to do with anything except that this is what the protein sequence of the spike matches up with. This is ~6 amino acids, which is admittedly not a whole lot, but not all furin cleavage sites look alike. If I am reading this blogpost correctly, the HIV-1 DNA sequence (which can represent the same protein sequence with worst case ~60% divergence) doesn't match up with the coronavirus, but the moderna patent sequence does (and it's exact). The moderna patent furin cleavage sequence known to be based on the HIV-1 sequence.
3. You ignore those results because NCBI will send you a hell of a lot of COV-19 sequences, because guess what, we've been sequencing the shit out of COV-19. But your criticism of the "statistics" that come out is 100% correct. It's possible that in the last two years have found a "missing link" that will be filtered out by those criteria, though one suspects that if such a thing were found, it would have made some pretty big headlines.
So, if we do a dispassionate analysis without considering faulty application statistics to the pre-2020-only sequence counts, and simply sticking to the existence or inexistence of sequences as per the database we are left with a few possibilities:
1.
COV-19 precursor protein intermixed with HIV-1 and did an uptake of the DNA (this is totally possible but rather rare, even given that HIV-1 is rampant in the wuhan-ish part of the world), then somehow converged on the moderna sequence (let's say 1 in 250 odds).
2.
Some (prior-to-2020 unknown or unregistered) DNA sequence that happens to exactly match up the moderna sequence, by chance, got merged with some wild Coronavirus sequence to yield covid-19.
3.
- someone came up with the idea of fusing furin into coronaviruses (true if you believe the leaked DARPA grant application, btw I have read the whole thing and the level of detail and its similitude to grants I have read and written would have been really hard, but not impossible, to fake, especially by a state actor.)
- someone went ahead and did the research on a slightly different coronavirus than proposed in the DARPA leak even without grant money (this does happen in biology, I'm probably on the tail end of the distribution but out of 10 years in science only 1 was working on previously proposed research, the other 9 were on research I was gathering preliminaries for to write grants)
- the postdoc or grad student assigned to this went and collected a bunch of furin cleavage sites, probably off of NCBI blast, and stack-overflow-copy-pasted them into the coronavirus, of which one was the site from the moderna patent.
- the moderna sequence randomly happened to be the "winner" and spread around.
4.
Moderna...
I've noticed that you are responding to many people that blatantly ridicule the original article and the author for writing behind a pseudonym. They are just trying to manipulate uninformed people towards ignoring this information.
The key to all this is that Coronaviruses have been studied for decades and there is no single coronavirus with that furin cleavage and HIV matches on the bindings.
After testing thousands of bats, no one has found an animal with COV-19 and, even worse, it is not easy to make bats to get infected by COV-19.
The fact that this short strings are in patents by one of the main manufacturers of the vaccines and the other ones are part of HIV which are directly connected to the Early Life of Fauci might be coincidental, but there is nothing that indicates that this is a natural virus.
Hold on. Let's be precise. Other coronaviruses do have furin cleavage sites. No other coronaviruses have this particular furin cleavage site.
> but there is nothing that indicates that this is a natural virus.
Let's also be clear here. You don't have to have anything that indicates that this is a natural virus; there is no reason why (as some of the other rabid commenters on the "other" side keep saying) this couldn't be natural. It would take some (very possible) stretches to get there. Biology does stretch sometime, there is no occam's razor, consider the platypus.
In the end I think one has to look at the proposed chain of events. My belief and let me be clear that this is a belief is that it's a lab leak. From when this started, I have half-jokingly said that this is how I would have mishandled the virus (I'm pretty clutzy in the lab - imploded no less than two dewars), and over time evidence has trickled in making the lab leak hypothesis stronger [0], while the observations you would expect have come out that makes the 'natural' hypothesis stronger have not surfaced, so in my mind, the trendline has been towards lab leak. Over time, the 'missing chain of events' that would have to have happened for it to be a lab leak are closing in. But sure, we could find a credible missing link for the natural hypothesis tomorrow and I would have to revise my beliefs. Honestly, the way the CPC has locked down information about what happened, we will almost certainly never know for sure and either way it will likely have to remain a belief.
[0] e.g. the leaked DARPA grant, information that the French refused to certify WIV lab, genetics on lao and yunnan viruses, similar lab leak happening in Taiwan, etc.
> It’s easy enough to change a single nucleotide (a single point mutation or SNP) or even insert or delete nucleotides (less common) but to insert 20 or 30 nucleotides with a code that works? Nope, that has to come from another virus or else it’s been done in a lab.
The original preprint which noted similarity of the HIV-1 sequences and those in the COVID spike protein, and on which this blog author's claims rest, was voluntarily withdrawn by its authors. [1]
Moreover, the assertions of uniqueness have been thoroughly addressed in the published article, "HIV-1 did not contribute to the 2019-nCoV genome" [2] :
> For any virus to obtain additional insert sequences from other organisms, it requires that it has direct interactions with other organisms, most likely through homologous or non-homologous recombination. [...] On the contrary, these motifs are widely present in various mammalian cells and so it will be more likely for bat CoV viruses to gain those motifs from the genomes of their infected cells if recombination indeed occurs.
That rebuttal article clearly suggests several feasible sources of the gene in Nature. That surely is more plausible than suggesting that Moderna or some other group nefariously created the COVID genome _in vitro_ before unleashing it upon humanity.
I deeply enjoy Hacker News for its ability to surface different ideas and sources of information that I would otherwise never find, but this article is simply ridiculous.
[0] https://en.wikipedia.org/wiki/Horizontal_gene_transfer
[1] https://www.biorxiv.org/content/10.1101/2020.01.30.927871v2
[2] https://www.ncbi.nlm.nih.gov/labs/pmc/articles/PMC7033698/
This is pretty common. The reason I take it as obligation is that if there aren't people fighting disinformationn, the disinformation will win, because it tends to be full of simple, seductive, but. wrong ideas that naive people fall for.
I'm not sure which particular thing public health people said that you're complaining about in the last part of the section- but I don't really think that public health leaders who communicate things which aren't 100% correct are dissemination disinformation. That's sort of a pedantic and uninteresting argument to have about the philosophical nature of information.
Even without spending time my heuristics is that I find biases in the claim and find out how referenced it is. This one had plenty red flags, including dismissing the RaTG13 results claiming that they were just put into the database after people started questioning, the adversarial tone, presumptions and so on. Conspiracy theorists use recognizable heuristics, but their motivation is to seek attention, for which there is large competition. They want to be relevant, but who doesn't. Conspiracy theorists are not really into publishing a paper either, a blog post, video or tweet is all you get, it's not about the truth and it is easier to avoid being found out this way, though social media these days have fact checkers, so they have to fly under the radar or keep to the social circles that are not bothered by fact checkers.
Knowing about biases and how to recognize them is a vital skill, should be taught in schools. Everyone needs a working bullshit meter.
To be fair, that sequence came from the Wuhan Institute of Virology
It is in the Moderna 2015 patent. 19 nucleotide match.
And so, fast vaccine production must be a priority. The development of drugs to treat viral infections and their manufacture must also be a priority. The ability to detect carriers and quickly shut down travel must also be a priority. And how do we enforce quarantine in populations that are resistant to it? All of these concerns are, I am certain, at the forefront in the mind of any Western government to be certain, but likely any government anywhere.
Because regardless of IF Covid-19 was created in a lab, it COULD HAVE BEEN created in a lab. And we all got caught with our pants down.
I propose that what we really see is that humans have a very hard time without someone to blame (hence victim blaming, hence religion and a whole bunch of other coping mechanisms).
I ain't a geneticist, so it goes without saying that I don't really know what I'm doing and I'm probably using BLAST wrong, but it seems to me like (EDIT: something very close to) CTCCTCGGCGGGCACGTAG is pretty dang common even within the human genome (let alone other species), and it doesn't seem far-fetched to me that SARS-CoV-2 might've yanked (EDIT: something very close to) that sequence from one of its hosts at some point.
EDIT: I forgot to check the completeness of the matches, and it does look like the hits tend to have one or two nucleotides that don't match. Still, it's close enough that the article doesn't seem like it presents much of a smoking gun.
¹: https://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Get&RID=Y3TZW9W...
²: https://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Get&RID=Y3UU86P...
After accounting for that, and after realizing that BLAST only returns the first 100 results, I searched on the "nt" database (where the article found "only" SARS-CoV-2 results) while excluding SARS-CoV-2 (taxid:2697049), "synthetic construct" (taxid:32630), and "other sequences" (taxid:28384). This¹ produced not just a lot of hits, but a lot of exact (i.e. 19/19) hits (mostly various microbes) - further suggesting that even the exact sequence in question ain't unique.
I also ran a search on "nt" strictly on SARS-CoV (i.e. the original SARS virus)², and while no single result completely matches the query, there are a lot of results that could be concatenated/combined to produce the query - and while I'm (again) no geneticist, it doesn't seem to me like some sort of impossibility for that concatenation to happen through mutations in the wild.
¹: https://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Get&RID=Y3WR4BP...
²: https://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Get&RID=Y3YCNK2...
Then you'd have Substack's talking about how the virus could have been naturally occurring, that there's no reason it had to be made in a lab, that it's just the government blaming China, etc.
I wish there was a word for beliefs that supposedly make use of evidence, but can easily use the exact same evidence to argue for the polar opposite belief.
Mutation? The same way we get every new pathogen?