6 ms·

> The first snakebite antivenom was made in the mid-1890s and the method has changed little since: snakes are “milked” for their venom, which is injected into h

by molf 2y ago

> The first snakebite antivenom was made in the mid-1890s and the method has changed little since: snakes are “milked” for their venom, which is injected into horses or sheep and the antibodies that their immune systems then produce are extracted via the animals’ blood. This is expensive, and many people develop allergies to the antibodies. A cool practical application of DeepMind's AlphaFold is that it allows one to "design" proteins because it is now possible to predict how they will fold. [0] This means it is possible to create synthetic protein antibodies that neutralise snake venom, specifically designed for humans [1]. There is a recent Veritasium video [2] that contains a great explanation. The whole video is worth a watch! [0]: https://ui.adsabs.harvard.edu/abs/2023Natur.620.1089W/abstract https://ui.adsabs.harvard.edu/abs/2023Natur.620.1089W/abstra... [1]: https://www.nature.com/articles/s41586-024-08393-x https://www.nature.com/articles/s41586-024-08393-x [2]: https://www.youtube.com/watch?v=P_fHJIYENdI&t=20m36s https://www.youtube.com/watch?v=P_fHJIYENdI&t=20m36s

mschuster91 2y ago
> This is expensive ... and the antivenoms have an expiration date, plus you'll need to keep the antivenoms distributed around the country at least enough to be able to fly it via a chopper anywhere it might be needed fast enough to be actually useful. The process for spider toxins is pretty similar with similar challenges. And on top of that, at least for snakes, even in a specific region the composition of what is in the snake venom might differ between local snake populations based on what the snakes eat.
perching_aix 2y ago
I did see Veritasium's video on this, and have been generally familiar with AlphaFold, but it remains unclear to me how the safety and efficacy of the antivenoms' designed with its help would be ensured short of a regular clinical trial gauntlet, which I understand continues to be extremely costly and long on its own. This isn't comparing it to the animal antibody harvesting approach, but on its own. Like I'm sure it's great and all, but I can definitely see it both mispredicting (mistakenly missing suitable proteins as well as mistakenly considering proteins suitable) and not having the predictive power to know how a given protein will interact with the countless atoms and molecules in the countless, somewhat arbitrary, random human bodies.
- mschuster91 2y ago
  > I did see Veritasium's video on this, and have been generally familiar with AlphaFold, but it remains unclear to me how the safety and efficacy of the antivenoms' designed with its help would be ensured short of a regular clinical trial gauntlet, which I understand continues to be extremely costly and long on its own. Well that one will have to be done for any medicine, the problem is snake and spider bites are rare in the Global North countries [1], but pretty prevalent in Global South countries. That means that there isn't much money in getting antivenins through certifications, there are dozens of majorly relevant species and hundreds of lesser importance which means a ton of expenses even if the antivenins share superstructures, and we'd run into the same issue as with the Covid vaccines - the accusation of local populations that they will be used as testbeds. [1] https://en.wikipedia.org/wiki/Epidemiology_of_snakebites https://en.wikipedia.org/wiki/Epidemiology_of_snakebites
- gus_massa 2y ago
  > and not having the predictive power to know how a given protein will interact with the countless atoms and molecules in the countless, somewhat arbitrary, random human bodies. You test it first in mice. If the new drug has not nasty side effects, you try it in other animals, then in monkeys, then voluntairs and then in pilot trial with humans.
  - perching_aix 2y ago
    > short of a regular clinical trial gauntlet
  - AStonesThrow 2y ago
    Nah you just make a Russian Roulette menu out of pharmaceutical offerings and continually parade new chemicals through the marketing machine as the patent granting/expiration cycle chews through them. You run costly ad campaigns and multimedia commercials, and you send salesmen out with imprinted pens and fancy notepads and they hand those things to the physician, then they purchase a nice lunch for the clinicians and explain how they treat all the weird nouveau conditions that were coined specifically for this drug. Do not Taunt Happy Fun Ball Then the insurance companies get on board with "fail-first step therapy" and the physicians start writing the cheapest prescriptions possible in hopes that the patient will go away or so totally fuckin' placebo'd that they were so busy filling scrips that they wouldn't bother complaining about the same thing twice. Can you tell that I immensely enjoyed Hank Azaria, Jake Gylenhaal, and Anne Hathaway in "Love & Other Drugs"?
j16sdiz 2y ago
AlphaFold helps the _design_ problem. ... but, How good are we at synthesizing the actual protein? Unless there are some breakthrough recently, Looks like we can't really do it at scale. Doing a few in lab setting is possible, but you can't really use them in drug.
- shiandow 2y ago
  In theory at least you could produce the mRNA and duplicate it and use it to produce the proteins. Though I'm probably missing a few steps here. A large part of this seems to be possible though, given the success of mRNA vaccines. Though using humans rather than animals to produce the actual protein is less than ideal.
  - afthonos 2y ago
    In theory, the difference between theory and practice is zero.
    - aaronmdjones 2y ago
      I've heard this phrased as "In theory, there's no difference between theory and practice. In practice, there is."
  - maxerickson 2y ago
    The methods for producing antibodies are well understood. There were several different monoclonal antibodies produced for use in treating Covid infections. I imagine the harder part is finding the most effective ones to manufacture, as it is still quite costly. Harvesting serum from a live animal has the benefit of harvesting whatever antibodies have been produced, rather than a single specific one, so an effective antivenom might need to be composed of a number of monoclonal antibodies.
- flobosg 2y ago
  > How good are we at synthesizing the actual protein? Quite good actually. Recombinant insulin is one example.
zosima 2y ago
There has been well-known processes for many years, to produce the relevant antibodies from industrial fermentation processes. If that is not being done, then it's surely because there is too little demand and the hurdles to get FDA to accept a new, obviously better serum, are too expensive to overcome.
pyrale 2y ago
Technology won't fix the issues described in the article, which include scam, african countries not receiving antivenom designed for local snakes, shortages, delay to get to treatment centers, and expensive products.
- noah_buddy 2y ago
  Maybe, maybe not. Cheapening costs can improve access, especially if it’s possible to perform a new cheap technique more widely.
- marcosdumay 2y ago
  It most likely will. A big reason why the antivenon for those snakes is rare is because one needs to keep a population of the snakes to produce them. If you don't need the snakes anymore, just the minimum amount countries have on hand for dealing with exotic fauna is enough to justify manufacturing them, and once you are manufacturing them, there's no reason not to do a larger batch and export some.
  - krisoft 2y ago
    I don’t understand what you are saying. Part of the problem is that the african country is buying snake venom specific to snakes in india. This is a problem which is not going to be solved by someone manufacturing a larger batch of snake venom and exporting some. In fact it is caused by that practice at least in part.
    - marcosdumay 2y ago
      Nobody is going to make antivenom for their snakes is large amounts because doing that would risk introducing their snakes on the local fauna. But if you optimize it for full in-vitro manufacturing, then anybody will be able to make it. And there's plenty of demand for somebody to do it, because lots of countries want small amounts of it.
- squigz 2y ago
  Why wouldn't it? > scam Probably not completely, but making the venom more accessible and affordable means less people are likely to fall for it and thus less people are likely to run a scam for it > african countries not receiving antivenom designed for local snakes A more abundant antivenom would mean more can be shipped to wherever it's needed most. > shortages, delay to get to treatment centers, and expensive products. Again, more abundance = easier to get, cheaper to make, etc.
flobosg 2y ago
> This means it is possible to create synthetic protein antibodies that neutralise snake venom, specifically designed for humans [1]. The designed protein does not necessarily have to be an antibody or nanobody, where only part of the protein is involved in antigen binding. The example you cite shows how smaller, single-domain proteins can also block and neutralize the venom.
refurb 2y ago
We are no where close to being able to accurately predict protein folding and the activity of the resulting molcules. Protein folding is incredibly complex. Imagine a string of 1,000 beads, each bead being made up of 6 different magnets all with the poles aligned in a different orientation. Now predict if you suspended it in zero gravity, how would it fold upon itself? Pretty complex? Well, it gets more complex. Now add millions of smaller magnets (water molcules) to the mix, each of which can attach to each magnet on the beads. Some will block the interactions that drive folding, others will enhance it. Pretty complex? It gets more complex. Now add a mixture of other magnets, some small (ions) some large (small molecules or other proteins) to the mix. Some have a single magnet pole, others have dozens. Now predict the structure and the resulting binding sites. AI can do a decent job of predicting the macro folding of proteins, but fails when it comes to accurately predicting all of the inter- and intra-molecular interactions. And if you can't predict all the interactions, you can't accurately predict what the actual activity of the molecule will be. At best, AI can highlight potential protein structures that might have activity.

by molf 2y ago

mschuster91 2y ago

> This is expensive ... and the antivenoms have an expiration date, plus you'll need to keep the antivenoms distributed around the country at least enough to be able to fly it via a chopper anywhere it might be needed fast enough to be actually useful. The process for spider toxins is pretty similar with similar challenges. And on top of that, at least for snakes, even in a specific region the composition of what is in the snake venom might differ between local snake populations based on what the snakes eat.

perching_aix 2y ago

I did see Veritasium's video on this, and have been generally familiar with AlphaFold, but it remains unclear to me how the safety and efficacy of the antivenoms' designed with its help would be ensured short of a regular clinical trial gauntlet, which I understand continues to be extremely costly and long on its own. This isn't comparing it to the animal antibody harvesting approach, but on its own. Like I'm sure it's great and all, but I can definitely see it both mispredicting (mistakenly missing suitable proteins as well as mistakenly considering proteins suitable) and not having the predictive power to know how a given protein will interact with the countless atoms and molecules in the countless, somewhat arbitrary, random human bodies.

mschuster91 2y ago
> I did see Veritasium's video on this, and have been generally familiar with AlphaFold, but it remains unclear to me how the safety and efficacy of the antivenoms' designed with its help would be ensured short of a regular clinical trial gauntlet, which I understand continues to be extremely costly and long on its own. Well that one will have to be done for any medicine, the problem is snake and spider bites are rare in the Global North countries [1], but pretty prevalent in Global South countries. That means that there isn't much money in getting antivenins through certifications, there are dozens of majorly relevant species and hundreds of lesser importance which means a ton of expenses even if the antivenins share superstructures, and we'd run into the same issue as with the Covid vaccines - the accusation of local populations that they will be used as testbeds. [1] https://en.wikipedia.org/wiki/Epidemiology_of_snakebites https://en.wikipedia.org/wiki/Epidemiology_of_snakebites
gus_massa 2y ago
> and not having the predictive power to know how a given protein will interact with the countless atoms and molecules in the countless, somewhat arbitrary, random human bodies. You test it first in mice. If the new drug has not nasty side effects, you try it in other animals, then in monkeys, then voluntairs and then in pilot trial with humans.
- perching_aix 2y ago
  > short of a regular clinical trial gauntlet
- AStonesThrow 2y ago
  Nah you just make a Russian Roulette menu out of pharmaceutical offerings and continually parade new chemicals through the marketing machine as the patent granting/expiration cycle chews through them. You run costly ad campaigns and multimedia commercials, and you send salesmen out with imprinted pens and fancy notepads and they hand those things to the physician, then they purchase a nice lunch for the clinicians and explain how they treat all the weird nouveau conditions that were coined specifically for this drug. Do not Taunt Happy Fun Ball Then the insurance companies get on board with "fail-first step therapy" and the physicians start writing the cheapest prescriptions possible in hopes that the patient will go away or so totally fuckin' placebo'd that they were so busy filling scrips that they wouldn't bother complaining about the same thing twice. Can you tell that I immensely enjoyed Hank Azaria, Jake Gylenhaal, and Anne Hathaway in "Love & Other Drugs"?

j16sdiz 2y ago

AlphaFold helps the _design_ problem. ... but, How good are we at synthesizing the actual protein? Unless there are some breakthrough recently, Looks like we can't really do it at scale. Doing a few in lab setting is possible, but you can't really use them in drug.

shiandow 2y ago
In theory at least you could produce the mRNA and duplicate it and use it to produce the proteins. Though I'm probably missing a few steps here. A large part of this seems to be possible though, given the success of mRNA vaccines. Though using humans rather than animals to produce the actual protein is less than ideal.
- afthonos 2y ago
  In theory, the difference between theory and practice is zero.
  - aaronmdjones 2y ago
    I've heard this phrased as "In theory, there's no difference between theory and practice. In practice, there is."
- maxerickson 2y ago
  The methods for producing antibodies are well understood. There were several different monoclonal antibodies produced for use in treating Covid infections. I imagine the harder part is finding the most effective ones to manufacture, as it is still quite costly. Harvesting serum from a live animal has the benefit of harvesting whatever antibodies have been produced, rather than a single specific one, so an effective antivenom might need to be composed of a number of monoclonal antibodies.
flobosg 2y ago
> How good are we at synthesizing the actual protein? Quite good actually. Recombinant insulin is one example.

zosima 2y ago

There has been well-known processes for many years, to produce the relevant antibodies from industrial fermentation processes. If that is not being done, then it's surely because there is too little demand and the hurdles to get FDA to accept a new, obviously better serum, are too expensive to overcome.

pyrale 2y ago

Technology won't fix the issues described in the article, which include scam, african countries not receiving antivenom designed for local snakes, shortages, delay to get to treatment centers, and expensive products.

noah_buddy 2y ago
Maybe, maybe not. Cheapening costs can improve access, especially if it’s possible to perform a new cheap technique more widely.
marcosdumay 2y ago
It most likely will. A big reason why the antivenon for those snakes is rare is because one needs to keep a population of the snakes to produce them. If you don't need the snakes anymore, just the minimum amount countries have on hand for dealing with exotic fauna is enough to justify manufacturing them, and once you are manufacturing them, there's no reason not to do a larger batch and export some.
- krisoft 2y ago
  I don’t understand what you are saying. Part of the problem is that the african country is buying snake venom specific to snakes in india. This is a problem which is not going to be solved by someone manufacturing a larger batch of snake venom and exporting some. In fact it is caused by that practice at least in part.
  - marcosdumay 2y ago
    Nobody is going to make antivenom for their snakes is large amounts because doing that would risk introducing their snakes on the local fauna. But if you optimize it for full in-vitro manufacturing, then anybody will be able to make it. And there's plenty of demand for somebody to do it, because lots of countries want small amounts of it.
squigz 2y ago
Why wouldn't it? > scam Probably not completely, but making the venom more accessible and affordable means less people are likely to fall for it and thus less people are likely to run a scam for it > african countries not receiving antivenom designed for local snakes A more abundant antivenom would mean more can be shipped to wherever it's needed most. > shortages, delay to get to treatment centers, and expensive products. Again, more abundance = easier to get, cheaper to make, etc.

flobosg 2y ago

> This means it is possible to create synthetic protein antibodies that neutralise snake venom, specifically designed for humans [1]. The designed protein does not necessarily have to be an antibody or nanobody, where only part of the protein is involved in antigen binding. The example you cite shows how smaller, single-domain proteins can also block and neutralize the venom.

refurb 2y ago

We are no where close to being able to accurately predict protein folding and the activity of the resulting molcules. Protein folding is incredibly complex. Imagine a string of 1,000 beads, each bead being made up of 6 different magnets all with the poles aligned in a different orientation. Now predict if you suspended it in zero gravity, how would it fold upon itself? Pretty complex? Well, it gets more complex. Now add millions of smaller magnets (water molcules) to the mix, each of which can attach to each magnet on the beads. Some will block the interactions that drive folding, others will enhance it. Pretty complex? It gets more complex. Now add a mixture of other magnets, some small (ions) some large (small molecules or other proteins) to the mix. Some have a single magnet pole, others have dozens. Now predict the structure and the resulting binding sites. AI can do a decent job of predicting the macro folding of proteins, but fails when it comes to accurately predicting all of the inter- and intra-molecular interactions. And if you can't predict all the interactions, you can't accurately predict what the actual activity of the molecule will be. At best, AI can highlight potential protein structures that might have activity.