What Adam is Reading - Week of 10-21-24

Week of October 21, 2024

 

We recently removed the backyard swing set from our childless house to convert the space into a Zen rock garden.  So far, our playground-to-rock garden conversion is beset with cost overruns, collateral damage (to trees, fences, and our irrigation system), and challenging landscapers.  To be sure, these are all first-world problems, but the philosophical irony is not lost on me.  As one of my friends reflected, "The trouble is you desired a Zen garden - an inherent violation of Buddhist tenets of non-attachment to material things."  Even backyard karma can be cruel.

 

-----

Here is the Google NotebookLM A.I.-generated podcast with two "hosts" discussing this week's newsletter.  For readers who would rather be listeners, check out What Adam is Reading, the unsettlingly good (and creepy) A.I. Podcast edition.

https://drive.google.com/file/d/1sQruoHhwGCR6_JmZKOBS30SCrjhYYSdu/view

About NotebookLM: https://blog.google/technology/ai/notebooklm-audio-overviews/

 

---

Based on wastewater concentrations, 1 in 111 Americans were shedding coronavirus in the last week (down from 1 in 30-40 in previous weeks).  It is worth reading some of the commentaries on historical and projected transmission -

https://pmc19.com/data/PMC_COVID_Forecast_Oct14_2024.pdf

 

The Pandemic Mitigation Collaborative (PMC) website uses wastewater levels to forecast 4-week predictions of COVID rates.

https://pmc19.com/data/

based upon https://biobot.io/data/

 

Wastewater Scan offers a multi-organism wastewater dashboard with an excellent visual display of individual treatment plant-level data.

https://data.wastewaterscan.org/

------

 

COVID articles

 

Scientific American published a fantastic, non-technical review of nasal vaccines on 10-15-24.   It highlights and reinforces why I am so eager to see these types of vaccines come to market.  To sum up quickly - Like injectable vaccines, nasal vaccines prepare the immune system for a fight but also offer additional protections - limiting transmission between individuals and decreasing the risk of infection after exposure to the virus.

https://www.scientificamerican.com/article/new-nasal-vaccines-offer-stronger-protection-from-covid-flu-and-more-no-needle-needed/

 

For those interested in a more technical article, Science published an excellent discussion about why the COVID vaccine has a hard time generating long-term protection.  The coronavirus spike proteins are too far apart on the viral membrane to fully active the immune system -  "B cells [the immune cells that are important for long-term immunity] carry Y-shaped receptors that attach to viral surface proteins when they identify a pathogen.  If both branches of the Y bind to the same pathogen proteins, they trigger a phenomenon called "cross-linking," which spurs B cells to transform into [long-term antibody-producing cells].  [The] SARS-CoV-2 spikes [on the surface of the virus] are about 25 nanometers apart, too [wide] for a single B cell receptor to readily bind to two at once [- thereby making longer-term immunity harder to generate]."  So, while short term responses to both the virus or vaccines work well, the longer-term immune “memory” is not as likely to form.   I wonder if this is an evolutionary adaptation the virus has undergone rather than an accident.

https://www.science.org/content/article/missing-immune-cells-may-explain-why-covid-19-vaccine-protection-quickly-wanes

 

 

Medical Trends and Technology

 

More data on GLP-1 drugs (like Ozempic).  A retrospective cohort study analyzing de-identified electronic health record data from the Oracle Cerner Real-World Data included 500,000 patients with a diagnosis of opioid use disorder and 800,000 patients with diagnoses of alcohol use disorder.  Patients with these diagnoses and who have GLP-1 prescriptions "appear to have lower rates of opioid overdose and alcohol intoxication.  The protective effects are consistent across various subgroups, including patients with comorbid type 2 diabetes and obesity."

https://onlinelibrary.wiley.com/doi/full/10.1111/add.16679

 

Deep-visual proteomics uses A.I. image analysis to scan the various proteins a cell produces.  European researchers recently mapped molecular changes associated with a life-threatening drug-induced skin reaction Toxic epidermal necrolysis (TEN) triggered in some patients when exposed by common medications.  Using deep visual proteomics, the researchers were able to identify protein pathways associated with TEN and find known medications that could block those pathways.  In both mouse models of TEN and seven humans with the reaction, the identified drug rapidly reverses the course of the disease (which is frequently fatal).  "Having identified the JAK/STAT (protein) pathway as an actionable therapeutic target in TEN using spatial proteomics, and given our extensive preclinical evidence, we treated seven patients with TEN syndrome, including three who were previously unresponsive to high-dose systemic corticosteroids.  All seven responded well to JAKi and were discharged in good health after treatment.  These data pave the way for a clinical trial of early JAKi therapy in TEN with the potential to change the outcome of this lethal adverse reaction to common drugs." The take-home message is that A.I. combined with biology is not theoretical - this combination yields actionable insights today.

Article:

https://www.nature.com/articles/s41586-024-08061-0

and

Good infographic from 2022 about deep visual proteomics:

https://www.nature.com/articles/s41587-022-01302-5/figures/1

from

Here is a summary discussion of the article from Eric Topol:

https://x.com/erictopol/status/1846570060980134340

 

 

Infographics

Our kids are back at college after a brief fall break.  I may have sent them the "College Degrees by their Return on Investment" infographic.  I may also have received feedback that I am not subtle.

https://www.visualcapitalist.com/which-college-degrees-have-the-greatest-return-on-investment/

 

 

Things I learned this week

 

Thanks to the New York Times, I now know the power and reach of Costco's monthly magazine.  This article offers a fantastic look into a part of American culture I did not appreciate.  (I should see if they are interested in nerdy nephrologists writing about science and tech.)

"Each month, 15.4 million copies of Costco Connection are mailed out to "executive" members, who pay double the yearly membership fee of $65 for the magazine and other perks.  (Another 300,000 are distributed via Costco warehouses.) Its reach is so vast that Costco Connection is now the nation's third-largest magazine by print circulation, behind AARP: The Magazine and The AARP Bulletin."

and

"If Costco often produces a kind of religious fervor among its shoppers, the Connection is its Scripture — a covenant and a recruiting tool that this month trumpeted the abiding allure of Lego bricks and ran a feature on Costco-themed Halloween costumes."

https://www.nytimes.com/2024/10/19/style/costco-connection-magazine.html?unlocked_article_code=1.Tk4.TrDh.7_mZU3t-IMjQ&smid=url-share

 

I learned that Australia has a version of National Geographic, aptly named Australian Geographic.  My introduction was a 2023 article entitled "Blood suckers and tear drinkers: The secret(ion) lives of butterflies and moths," describing the behaviors of these insects to ingest salt and other minerals through the tears and blood of different animals (such as turtles and alligators).  While the article is clear that this behavior is observed in South America, it is fitting for an Australian magazine to discuss vampire-like butterflies.

https://www.australiangeographic.com.au/topics/wildlife/2023/08/blood-suckers-and-tear-drinkers-the-secretion-lives-of-butterflies-and-moths/

 

 

Living with A.I.

 

Sequoia Capital published a thoughtful white paper on the future of generative A.I.

https://www.sequoiacap.com/article/generative-ais-act-o1/

 

Here is more on the intersection of genomics, proteomics, and A.I. - a multimodal A.I. that allows users to explore the data from any single-cell RNA-sequence analysis via natural language conversation.  No coding is needed.  This is one of many tools in development, but the

I highly recommend watching the 2.5-minute video at

https://cellwhisperer.bocklab.org/

https://x.com/BockLab/status/1847204272565354675

Play with the tool: https://cellwhisperer.bocklab.org/geo/

 

"The Department of Defense wants technology to fabricate online personas that are indistinguishable from real people."  Though this is, essentially, an article about a DOD RFP, the strategy behind the RFP is quite striking.  Moreover, the ethics and prevalence of Deep Fake personas (think Russia and China) used to sway public opinion is well worth debating.

https://theintercept.com/2024/10/17/pentagon-ai-deepfake-internet-users/

 

 

A.I. art of the week (A visual mashup of topics from the newsletter).

A butterfly with vampire fangs is sitting in a lotus position on a stone bench in a Zen rock garden.  It holds a Costco membership card, and in front of the bench are numerous magazines with the titles Connections and AARP.

https://drive.google.com/file/d/1RTpubggHrbvZbT4k3ANs6hqtfyH1tQcT/view?usp=sharing

 

 

 

Clean hands and sharp minds, team

 

Adam

Comments