> For the first half of the twentieth century, the going theory was that the invading element—the “antigen”—served as a template around which a corresponding antibody was molded. Only in 1955 did scientists discover the much stranger truth. It turned out that the cells that produce antibodies—called B cells, because they were first discovered in the bursa of Fabricius, an organ that does for birds what bone marrow does for humans—can produce only one kind each. Its structure is random, and nearly every B cell is discarded unused. If, however, an antibody created by a B cell happens to match some part of an antigen, that B cell will not just survive but clone itself. The clone incorporates many mutations, which offer the possibility of an even better match. After a few generations, an antibody with the best fit is “constructed” through a process of mini-evolution that occurs continuously in our lymph nodes and spleen. (Our ancestors the bony fish adapted the machinery of the B-cell system from an even more ancient parasite.)
So, natural selection discovered a way to run an embedded natural selection search process in an individual. Talk about meta.
After learning CS I can’t help but see biology as crazy algorithms run on a living computer. Like someone was thinking “how can we match possibly any shape of invader? Screw it cells are cheap we’ll brute force randomly try them all.”
Tons of biology can be modeled mathematically strikingly well, and most of that was figured out a long time ago. Sewall Wright was developing graphical models of evolution in the 1920s.
It’s mind blowing not only how our body mimics natural selection with B cells, but also how it generates the genetically diverse B cells in the first place. [1]
Yes, and if you have different versions of the HLA gene, you get different distributions as a result of V(D)J recombination.
So, in a way, different individuals have different immune strategies. For example, super-responders to HIV tend to be HLA-B57. This HLA "version" trades better response to virus that mutate quickly for more chances of autoimmunity as thymic selection is less stringent (because HLA-B57 presents less selfpeptides in the thymus).
Culture, intelligence, memes, innovation etc. are also evolutionary subprocesses of our species. Evolution is gradient descent, so is learning. It's all the same. Entropy minimization or competitive negentropy consumption, if you want.
Our brain also produces hypotheses, tests them and repurposes the most successful ones as program fragments stored in synaptic weights. Trial and error, exploration, mutation, survival of the best hypotheses etc.
extremely naive, dumb questeion: could cancer cells stem from some errant mutation of B cells? this might help explain why cancer cells grow uncontrollably and also escape detection by the immune system.
Also the replication machinery being easy to trigger is why it's the second most common cancer after skin cancers. (Which shares some of those features in addition to being even more exposed to insults and less controlled.)
Where do you get that figure from? All the figures I have state lung and bowel cancers are the most common, along with prostrate and breast for sex specific cancers (although looking at their rates they would be the higher even though their rates in the other sex are exceedingly low).
That's because melanoma and non-melanoma skin cancer are split, as other lymphoma and non-Hodgkin lymphoma, multiple myeloma etc. in data despite affecting same cell lines.
If you sum these two up they end on top respectively in that order.
Albeit prostate cancer is really close.
It's an artifact of grouping used. For colorectal for example, adenomas and carcinomas are lumped for a change.
Prostate is the second most common worldwide, or skin cancer, depending on year. The big 5 are over 1.5 million. (WCRF data. They lump all lung cancers but split immune system and skin cancers. For reasons? Bladder and lung cancer incidence is highly dependent on factors such as smoking and specific chemicals used in manufacturing. Potentially the same with colorectal and diet. Rates for breast and prostate are high because they're easy to spot and well tested. So per country data varies a lot.)
phrased another way: could the same process that spawns these legions of B cells also inadvertently spawn cancer cells?
related: healthy cells exhibit properties that allow them to go ignored by the immune system, similar to a VIP pass for bypassing security at a basketball game. is this right? if yes, what are the minimal properties that constitute this cellular VIP pass?
B cells carry the same DNA as all other cells. The difference between cell types is in what sections are functionally expressed (made into proteins that do their work). This is determined by a multitude of factors.
Cancer cells typically have mutations, even just single base pair changes, which change the way that genes are expressed in the cell and thereby influence such behavior.
The original functions of the B cell aren't necessarily needed for other cell types to exhibit similar functions. The base plan is there, it just needs a nudge.
There are already examples of alternatives, like it could have some kind of brain and communication apparatus to work with others of its species and put together things like GPU simulations of protein modeling to generate optimized artificial antibodies.
I believe what is meant here is that the alternative is to have a highly developed capacity for abstract thought that you then use to collaborate with your species to develop antibodies in a laboratory (humans are an example of this, although we don’t rely upon this mechanism)
Yeah but how is that going to arise out of nature? It's like asking why birds use vision + magnetism to navigate when they could just use GPS -- the MVP is too massive and inflexible.
I am blown away by this casually dropped metaphor:
>As the virus spreads unchecked through the body, it drags a destructive immune reaction behind it. Individuals with COVID-19 face the same challenge as nations during the pandemic: if they can’t contain small sites of infection early—so that a targeted response can root them out—they end up mounting interventions so large that the shock inflicts its own damage.
It should also be added that an out of control infection is not the only reason why you end up with an immune response that harms the host. It's quite possible to get a severe immune response to an infection that's quite limited.
Kind of similar to anaphylaxis to bee stings. The venom was never any threat to a human, but the immune response sure can be.
Really like this analogy: It is a somewhat astonishing fact of life that the exact same DNA is shared by every cell in your body, from the skin to the brain; those cells differ in appearance and function because, in each of them, a molecular gizmo “transcribes” some DNA segments rather than others into molecules of single-stranded RNA. These bits of RNA are in turn used as the blueprints for proteins, the molecular machines that do most of a cell’s work. If DNA is your phone’s home screen, then transcription is like tapping an icon. By sampling the RNA present in a group of cells, researchers can see which programs those cells are running at that moment; by sampling it after the cells have been infected with a virus, they can see how that virus substitutes its own software.
'Molecular machines' is not an overstatement. Here's an example of some of them working in harmony to maintain an essential electrochemical gradient across the lipid bilayer that encapsulates the mitochondria:
It's a 7 minute video, if you only have a little time, skip around, but make sure you listen to the part starting at 6:17.
DNA isn't just the blueprint for the machine or the blueprint for the factory that makes the machine, it's the blueprint for the entire supply chain required to build and operate an entire chemical industry of mind-bending complexity. Like doubling grains of rice on each square of a checkerboard, if you wanted to create a complete catalog of all possible combinations of a string of DNA that's 150 base pairs long, you would have to marshal all of the estimated matter in the observable universe and you'd still end up short. Most of the genes that create these proteins are tens of thousands of base pairs in length and the entire human genome is ~3 billion.
Thanks for posting this. Looked further into atp synthase. Absolutely amazing. A little turbine, driven by the flow of protons, spinning in 120 degree discrete clicks, around catalyst pockets, where each click changes the shape of the pocket to a new one. And after the pocket has gone through a complete cycle, adp and phosphate have been fused.
Never seen the chemical (proton concentration), converted into something mechanical (spinning), and back into chemical (adp + phosphate -> atp) in such a way before.
ATP synthase is amazing. It uses the driving force from the potential gradient (aka voltage) built by the proton pump in the first video to spin at ~21,000 rpm and crank out ATP. I read that your body creates (and then consumes) about its own weight in ATP every day.
When I think of these systematically I just wonder what the intermediate steps were. The proton pump uses energy to create the potential gradient that ATP synthase runs on. You can't have a proton gradient without the proton pump, and without the proton gradient you don't have ATP. So these two exquisitely complex machines depend on each other and had to evolve together. It boggles my mind.
I feel like learned nothing about how the pump works from that sodium - potassium video. They just handwaved it away with "conformational changes". Would be neat if they zoomed in on the moving parts.
Immediate step is a cell which is long and has a natural ion gradient. E.g. one with flagella on just one side. Then the better ones would evolve machinery to speed up the gradient, which happened to be using some energy but made said cell move and digest faster, thus also reproduce faster. At this point this could still be shut down during lack of food or activity, as there was a natural gradient when activity was nearby.
The membrane is the later invention, dating to organelles which are suspected to be microcellular parasites that got integrated and then reduced.
These membranes were much more efficient, however they were very hard to turn off for hibernation as the membrane is a static gradient and hard to disrupt temporarily.
So if you have a long cell with an ion-gradient along it, I guess you have to have the turbines attached to something right? Or else the protons will just push it around rather than spin it? And then there is still the issue of protons just flowing around the turbine rather than through it. It may be possible, but to my layman intuition it sounds way harder to get right than a membrane.
I wonder if maybe an intermediate step could be just channels and turbine. But no pump. If the amount of protons surrounding membrane varies over time, the cell can open the channels when there are many protons outside, and close the channels when there are few (forcing equalization through turbines).
Or maybe it can use acid protolysis to create protons?
Good analogy, but there are some differences - your cells are not running one app, but just tens of thousands of routines with no clear definition of which routine is doing what function (many do multiple functions often with mix-match redundancy). And the mere fact that a routine is running is not enough, their activation state (posttranslational modifications, folding configurations, binding partners) or the values of all its internal variables is also important to know the exact function of each routine.
Even that is only theoretical - we generally know some level of detail to know what many of these routines are approximately up to, but rarely in the detail needed to figure out what a cell is up to in general.
In some ways studying what a cell is upto using our current methods is comparable in certainty to how we poll the US before elections.
The really mind blowing thing is not just that DNA encodes everything for a living organism, but that it also has all the necessary elements to control expression so the right thing gets made at the right time in the right place.
The fun part is that it's not like DNA is just expressed, it needs to be "unwound" from it's natural state as a highly dense structure. Elements within the cell actually regulate the unwinding itself, that's one way genes are turn on/off.
And in addition, once DNA is unwound, split into a single strand and transcribed into RNA, the RNA itself becomes a regulating mechanism through siRNA. So the very act of transcribing DNA has regulating effect up the transcribing itself.
Conjures up a mental image of a computer where the only mutable state is instruction pointers for parallel threads. Not quite accurate, but a rough ballpark to how crazy those "programs" must be.
(If for some reason we had to work with that kind of computer we'd probably enforce a conventional, graspable approach on it by finding a way to simulate a conventional computer once and then conveniently forget about the underlying madness, clearly that wouldn't be the path taken by evolution)
If you're looking for the summary, it's about 1/4 the way in.
> Usually, the viruses that humans care about are successful because they shut down both of these signalling programs. The coronavirus is different. “It seems to block only one of those two arms,” tenOever told me. It inhibits the interferon response but does nothing about the cytokines; it evades the local defenses but allows the cells it infects to call for reinforcements. White blood cells are powerful weapons: they arrive on an inflammatory tide, destroying cells on every side, clogging up passages with the wreckage. They are meant to be used selectively, on invaders that have been contained in a small area. With the coronavirus, they are deployed too widely—a carpet bombing, rather than a surgical strike. As they do their work, inflammation distends the lungs, and debris fills them like a fog.
I've been doing some lab & modeling research on this, and it's more complex. I think it's a non-linear dynamics problem.
In a nutshell, SARS-CoV-2 (COVID-19) does block interferon response by using molecular mimics. Some of the proteins encoded in the virus are really similar to proteins in the interferon pathway, so this is not surprising. Still pretty unexplored and a nice route towards finding treatments.
Now, the interesting part is that the dynamics of interferon is very non-linear. Cells exchange interferon signals with other cells in the same tissue to pass information and coordinate with the immune system. There are fairly sophisticated models (using stochastic message passing!) that confirm lab observations on regular non-infected cells [1].
My hypothesis is that by blocking interferon in a sufficient number of cells on individuals with already disrupted innate immunity (due to ageing, insulin resistance, etc.) the system becomes chaotic. Infection by e.g. pneumococci is tolerated because the signaling is perturbed and then, all of a sudden, you get a massive amount of proinflammatory molecules secreted and cells decide to undergo apoptosis. That's the cytokine storm that everyone talks about.
Sadly, pursuing these ideas within academia is hard as most (but not all) biology labs are really hostile to mathematical models, even if they explain or predict things they can't using hand waving. Perhaps I should seek VC funding.
bio people are not frequently math people and vice versa. The undergraduate departments make fun of each other. This is changing slowly. I suspect that a big part of the reason is that biology is complicated while math is good at dealing with distilling the behavior of a few essential elements.
When you start to find essential elements, they get mathified (look at DNA) but look how much experimental work it takes to do that? Scaling up experiments can also take some math, but once you figure out the trick, you just have a lot of experiments to run....
And this is why steroids like decadron are helpful in the second phase (now being called the inflammatory phase by many in the ID community) of the infection. Steroids decrease inflammation / immune response.
That's not exactly why it's steroids. If this was the main mechanism we'd use more effective immunosuppressants. As is, stronger immunosuppressants show worse survival rates. You want to keep some immune system activity to not get a secondary infection and fight the virus.
Corticosteroids also kickstart healing process (not really in COVID) and regrowth - you also race against tissue destruction. Most importantly they barely reduce scar formation. And they actually improve breathing via action on alveoli as well as reduce swelling.
The specific two to use are hydrocortisone and dexamethasone based on studies thus far, in critical patients - ones put on respirators. (Prednisolone could also work.)
This would be the proper description for the spanish flu. But with Covid-19 it's different. No cykotine, but BK overreaction. Healthy immune systems are not triggered to kill the body, only weak and old are in danger. Besides the neurological damage in some.
This kind of reporting interleaved with storytelling in a really weird voice is frustrating. Why do journalists write like this? What purpose does this word salad serve?
I can do this myself! "On a bald September day, a man walks into a barber shop, languishes at the occupied seats, and voilà, while sensing upcoming bowel movements, he discovers the next great theory of economics."
Cosmos was written in a similar style. Try reading it in Carl Sagan's voice and see if it's any more interesting to you.
The purpose of longform writing is to provide flavour and context, which are important parts of an information diet. Not everything has to be about the freshest facts delivered to your brain as urgently as possible. Just because you've happened across this in a frenzied binge of clicking HN headlines, doesn't mean it was designed to be consumed that way. If you don't feel you have time to sit through a New Yorker article, save it for later.
This is 'Long Form Journalism'. The 'dip in, dip out' paragraph structure is a dead giveaway.
Not always what you're looking for but it can be a good way of cracking into a subject you're unfamiliar with, or a good way of merging the human element with something that would otherwise be quite dry.
By no means a substitute for regular news, but a good complement to it.
Related to unnecessary words inside a newspaper/magazine article, this interview [1] with Georges Simenon is the best example of why those articles look like crap (at least for people like me, I do realize that there are other people with different tastes):
> Just one piece of general advice from a writer has been very useful to me. It was from Colette. I was writing short stories for Le Matin, and Colette was literary editor at that time. I remember I gave her two short stories and she returned them and I tried again and tried again. Finally she said, “Look, it is too literary, always too literary.” So I followed her advice. It’s what I do when I write, the main job when I rewrite
> I: What do you cut out, certain kinds of words?
> Adjectives, adverbs, and every word which is there just to make an effect. Every sentence which is there just for the sentence. You know, you have a beautiful sentence—cut it. Every time I find such a thing in one of my novels it is to be cut.
I think it is actually optimized for print magazines where someone can be expected to dip in and out when they have a few minutes on the couch and the publisher has a set number of pages to fill. It's not as well suited to the internet where 25 words surrounded by 10 ads seems to be the norm and 5 minutes is an eternity.
I don't disagree but national geographic, time, and now the same formats transposed into the online world all have their place. They can act as fantastic springboards into underlying news, scientific research, and other further reading in a way that's similar to but far more engaging than a dry wikipedia article.
You can condense the facts into 3 paragraphs but why don't we just go all in and present them as bullet points instead of full sentences?
When I see a five page article, I know that I'll get background information and I might discover something I didn't expect from the title. While you might only be looking for the study results, I like spending my Sunday afternoon reading long articles teaching me all kinds of interesting information I wouldn't have looked up on my own.
And what a shallow, frenetic dystopia it would be if nobody even attempted to write compelling longform anymore. Remember, 90% of everything is crap. Imagine reading 3 disappointing novels and concluding that novels, as a format, were a waste of time!
(Or worse - being unable to appreciate a good novel because you're too impatient for it to get to the exciting bits!)
Compelling means that the reader doesn’t give up after a few paragraphs. I like long form, like I have subscribed to Harper’s since forever, but sometimes an article is just extended and not more interesting.
I find NPR guilty of the same. I want to hear a story about how fear affects the lives of primates. And it begins on a cool October day with thoughts of coffee.
I suspect that it's a holdover from the past. There once was a time when many people had more time to read than they had stuff to read. This kind of writing was fine. Better than right-to-the-point writing, in fact, because it expanded the amount to be read without actually needing to have more to say.
Alternately, it's a way of expanding readership. If you don't know anything about molecular biology, they have to write an article at a layperson's level to explain what they mean. But if you, the reader, don't want to go through the effort of trying to understand even that, then there's this human-interest stuff that may still cause you to decide to buy their magazine.
I'd wholeheartedly agree with your critique if the text was published as a tweet storm.
The thing is, this isn't news reporting at all. It's a popsci writeup of the history of the immune system, the history of the discovery of the immune system and a glimpse into its complexity that uses Covid news as the textual equivalent of a cover image. If you have a print of 20000 Leagues under the Sea and the cover shows the kraken attack, would you complain about the number of chapters that don't contain a kraken at all?
It's because students are being admitted to universities (including ivy league ones) based on informal political quotas. Once inside they don't learn per se but are indoctrinated with the virtues of diversity and tolerance. Nobody should be amazed that when they graduate, they write like a 12 year old: that was the whole point.
It would be interesting to see a heat map for reading an article like this.
It’s so long.. after 10 minutes I wanted to know the point of it, so I skipped aggressively to the middle, found a few points, then skipped to the end.
I wonder if there’s any value to so many words in this day and age with so many other things to read on the internet.
Every time a New Yorker article shows up, people crawl out of the woodwork to complain that it's a New Yorker article.
If New Yorker articles are not for you, then do not try to read New Yorker articles. Really! Read other things you will actually like. If you want a quick rundown of the topic, Wikipedia is only a click away, and gets right to the point.
Those of us who like New Yorker articles appreciate them when they appear. We know there will be an investment of time, and we know that a board of Editors has ensured that, for those of us who like the format, our time will not be wasted. Those who don't, know in advance that they will just be frustrated and bored, and can skip past instead. You are not even obliged to "upvote" the post.
Really, nobody is keeping track. Nobody will think ill of you for skipping past the moment you notice "New Yorker" in the URL. If you ever become ready for New Yorker articles, New Yorker articles will be there for you.
When you post an article, what sort of discussion are you hoping people will hold?
If you are hoping that people will discuss the novel aspects of the coronavirus's impact on the immune system then you might choose an article that focuses on that.
Instead, if you choose an article that spends a large amount of time discussing tangential points, creating lengthy analogies for the workings of the immune system and discussing other forms of immune response then probably you are hoping people will discuss those too.
But people will appreciate as well as criticise. That there are people who appreciate the analogies are apparent in the discussion here where people do just that. That there are people who would prefer a focused article and find that the tangents detract from the underlying store is just the flip side of that discussion that the op decided to have.
I personally wish that long form journalist articles would have an outline at the beginning so that I know what I am about to read. Otherwise it often just feels like the author is rambling and going nowhere. Some stories actually have a lot of moving parts and require long form journalism to paint a realistic and balanced picture. Instead, a lot of "long form" journalism just adds filler analogies and tangential factoids that may be interesting but often don't add anything to the underlying story or are unnecessary for understanding.
I can tell you what we don't hope to read: complaints about New Yorker editorial policy, or about long form journalism in general. Again: it has what it says on the tin. We get that you don't care for it. You are absolutely not obliged to post anything at all.
Another related mechanism is antibody-dependent enhancement (ADE)[1], where suboptimal antibodies bind to the virus and cell receptors, promoting the former's entry into the latter. It has been proposed that ADE increases the severity of SARS-CoV-2 infections[2].
Worst part of the way these policies are deployed is that they rarely take into account how people will react when forced to solve their issue rapidly... they'll go ask many people .. more infections.
Thankfully, we do have a (now scientifically proven !) working "cure" – though it mostly works at the onset of the first symptoms, when the virus is only starting to get a foothold in the body :
Ok, that doesn't look good…
(Still, I would have preferred that doctors would be free to decide by themselves which drug to prescribe for a new disease. After all, they have skin in the game !)
> For the first half of the twentieth century, the going theory was that the invading element—the “antigen”—served as a template around which a corresponding antibody was molded. Only in 1955 did scientists discover the much stranger truth. It turned out that the cells that produce antibodies—called B cells, because they were first discovered in the bursa of Fabricius, an organ that does for birds what bone marrow does for humans—can produce only one kind each. Its structure is random, and nearly every B cell is discarded unused. If, however, an antibody created by a B cell happens to match some part of an antigen, that B cell will not just survive but clone itself. The clone incorporates many mutations, which offer the possibility of an even better match. After a few generations, an antibody with the best fit is “constructed” through a process of mini-evolution that occurs continuously in our lymph nodes and spleen. (Our ancestors the bony fish adapted the machinery of the B-cell system from an even more ancient parasite.)
So, natural selection discovered a way to run an embedded natural selection search process in an individual. Talk about meta.