[gtmb]

If we now have (1) a better understanding of proteins (i.e. AlphaFold) (2) studies like this that increase our understanding how one microbe attacks another

(1) Couldnt we create proteins that are able to kill a specific microbes? (2) Put these proteins in a "dialisys" machine, so that as our blood passes thru the machine, the proteins there kill the microbe (3) Same protein or other protein help the machine separate the debris of the dead microbe from the good blood

How feasible would that be with the technology we have? To me it seems a good idea, as focusing on this type treatment drug development would be much shorter - since the drug/protein never enters the body - so you wouldnt need long clinical trials.

And given the rise of super resistant bacteria, also seems a good idea to my non-specialist self.

[woke_neoliberal]

There are some proteins that directly cause lysis (called "lysins"), e.g. there are lysins taken from bacteriophage (viruses that attack bacteria and are generally species or strain specific) that do kill bacteria effectively.

One major issue with chemical antibiotics (proteins or small molecule) is evolution. Antibiotic resistance is largely the product of the wide and incomplete treatment of bacterial infections, and the resulting emergence of resistant strains. Another issue, shy of evolution of genome modification, is simple gene / transcriptional regulation, and up-/down-regulating surface proteins and sugars that can render such proteins ineffective. It's not as simple as binding a protein target.

Consider that resistance to a novel antibiotic has been observed after a year or two on the market, and it doesn't make sense for a pharma company to invest in bringing one to market.

One interesting avenue is using bacteriophage to treat infections directly. The phage evolve with the bacteria and have a much higher chance of remaining effective. Treating with an antibiotic, or multiple phage, using different mechanisms, decreases the chance of evolving resistance. Phage have been used in eastern European countries for ~ a century, but were dropped in western countries in favor of small molecule antibiotics. There are several clinical trials to bring them back to market.

To bring it back to the article, interestingly phage don't seem to be mentioned, even though they are the most abundance organism on the planet. Surely something has figured out how to eat them...

[emeraldd]

There are a few problems, beyond the one in the other comments that come to mind. (1) How are you assured to keep the proteins in the machine? (2) could this lead to a prion infection? My understanding of prions is limited so take this thought with a grain of salt. I was under the impression that while there may not be anything different with the composition of a prion pathogen, they tend to be misfolded and that misfolding tends to be contagious to other properly folded proteins.

[rvba]

I wonder if there isnt something that eats prions too. After all, if prions were impossible to destroy, then the whole planet would be one big pile of prions.

Didnt the planet have a LOT of coal since for millions of years there were no fungi to digest the dead trees?

[tremon]

We probably could. But the prevalent wisdom in pharmacology is that the hard part is not killing the thing you want killed: the hard part is keeping the healthy tissue alive. And I don't think we have adequate models (nor the processing power) to evaluate every potential cure protein against every potential cell in the human body. But that's just a guess, it's not my field of study.

[carlmr]

Didn't they have this AI to predict the toxicity of certain compounds with extreme accuracy?

If you combine the two it sounds like a solution with some remaining risk, but if the disease is bad enough that risk may be worth it.

[colechristensen]

Oof, people tend to way overestimate the capabilities of these tools.

No there are not magic AI tools where you can wave your hands and design proteins or evaluate the properties of arbitrary compounds.

[fbdab103]

It is difficult for the layman to appreciate when every other day there is a puff piece about how group x has solved protein folding or what have you.

[fbdab103]

Consider me highly skeptical of any AI models predicting biology.

The way these models were is more or less by association. There is no foundational understanding of underlying pathways or likely decomposition products. The AI models can be used as a first stage filter: generate a bunch of putative products computationally, filter them through predicted toxicity profile, and then start real testing (though you are likely leaving out solid leads from false positives).

AI can be a human assisting tool, but it is in no way sufficient for leap-frogging into human testing.

[svnt]

If I understand you I think something like this is what you are talking about, except this is perhaps better because it doesn’t require the blood filtering:

https://phys.org/news/2019-12-protein-medicine.html