Hacker News Reader

Comment by deepzn | Hacker News Reader

With new drugs released in the past year for Alzheimers after more than 20 years. It's exciting that we may be beginning to understand this disease.

Also excited to see how LLM's and DL/AI can help accelerate research by reducing menial tasks for researchers and scientists as well as by contributing to drug discovery. https://medicalxpress.com/news/2023-05-scientists-ai-drug-al...

View on HN

Which drug are you thinking of? Because the recent one that was approved was both highly controversial because the efficacy is extremely questionable[1] and the entire theory of disease is in question as well [2].

[1]: https://www.npr.org/2021/06/07/1003964235/fda-approves-contr... [2]: https://www.nbcnews.com/science/science-news/alzheimers-theo...

They may be thinking of the one approved even more recently [0], of which I am unaware of any controversy.

[0] https://www.theguardian.com/society/2023/may/03/new-alzheime...

Yes, this too. This one though appears to have some side effects of brain swelling as mentioned in the article.

Though the one announced in 2021 made the headlines and appeared to be more of a success than it was. You could say a new drug being approved after 20 years is still a relative success (it does have some statistically significant results which led it to being approved).

The beta-amyloid debacle will persist as one of the most fatal mistake the scientific community, and a single individual's deceit created.

If you're referring to the (likely) falsified research of Sylvain Lesné, my understanding is that researchers universally agree that it was not central to the beta-amyloid hypothesis, and therefore didn't change much. The story appears to have been widely mischaracterized in the press, even by Science, who broke the story.

> and the entire theory of disease is in question as well [2].

The entire theory of Alzheimer's is not resting on that research. Nobody believes in amyloids exclusively because of it.

Yes these anti-amyloid drugs announced recently. I wasn't aware of the controversy. Hopefully, this period is the start of more fruitful discoveries.

These new drugs, or something else? https://www.science.org/content/blog-post/are-anti-amyloid-a...

I didn't read earlier about the backlash, but yes, I was talking about these-https://www.bbc.com/news/health-63749586

But I see there are side effects as well (Thanks for pointing it out) -https://www.axios.com/2023/05/15/alzheimers-drugs-patients-r... -https://www.science.org/content/article/scientists-tie-third...

It's good then that we now have another solution to tackle this like mentioned in OP, such as targeting reelin or APOE. And if anti-amyloids, or reducing amyloids are only a part solution. Hopefully, the findings keep continuing to cure Alzheimers.

I read a little more In The Pipeline than I should, and understand a little less of it than I should.

How far are we from being able to simulate every molecule and folding in our body, tell a neural net what the desire outcome is and have it point out everything that prevents the desired outcome and what to do about it in a way that also doesnt prevent the desired outcome

Here's an interesting paper that looks at blood plasma protein contents and uses bioinformatics processes including learned models to identify plausible biofeedback pathways responsible for protein concentrations deviated from baseline. It's not what you are asking for, but it's the closest thing I've seen to date:

Plasma Proteome of Long-covid Patients Indicates Hypoxia-mediated Vasculo-proliferative Disease With Impact on Brain and Heart Function (Preprint)

https://assets.researchsquare.com/files/rs-2448315/v1/8043bd...

An excerpt:

> In Fig. 7A, hierarchical clustering heatmaps reflect the levels of neurological markers across the patient groups (markers have been curated by OLINK). The values of the PEA expression levels were hierarchically clustered based on Pearson correlation algorithms. Markers selected through the above methodology were investigated for functional annotation using tools from the GSEA platform and MSigDB data positories (Fig. 7B). This latest analysis demonstrated that functional clusters were formed around leukocyte migration, positive immune signals, glial cell differentiation, neurogenesis and MAPK regulatory modules. Taken together, these pathways predict a possible brain-blood barrier dysfunctionality grounded on cell proliferation. Graphs in Fig. 7C illustrate the expression levels of individual markers from the functional groups presented in Fig. 7B. One of the highly expressed markers, was the amyloid precursor protein (APP; Supplementary Fig. 10) which is known to be a pathognomonic marker for both Alzheimer disease and brain inflammation [61–65]. Additional markers for brain dysfunction include JAM2 (endothelial tight junctions protein), SNAPIN (a mediator of neuronal autophagy-lysosomal function in developing neurons), KCNH2 (potassium channel), S100A14 (involved in cell motility adhesion and growth), KIAA0319 (language impairment biomarker), and IROR1 (a receptor tyrosine kinase like orphan receptor 1, which regulates neurites growth in the central nervous system having also WNT-signaling pathway functions, and being crucial for the auditive apparatus maintenance).

Incredibly far. Simulating all of that would be egregiously wasteful and extremely unlikely to provide actionable results.

I'd guess about 0.1% complete.

Still a ways off but AlphaFold is a big step in that direction.

Reply on news.ycombinator.com