Fediverse

@stealthradek I don't think that's yet on the roadmap (and I didn't find an issue describing the idea at https://codeberg.org/comaps/comaps/issues yet)

If you want, you could open an issue suggesting it, but as you mention, it's quite a rare use case. I fear, unless someone with the right skill set to add this runs into the problem themselves, it seems unlikely it'd be prioritised soon.

@inkeduprunner Super simple

🧬 Scientists are using mRNA to make cancer cells produce an immune alarm protein called IL-12, and early results are striking.

The idea: inject mRNA encoding IL-12 directly into tumors. The cancer cells produce the protein, which activates T cells and NK cells to attack. Systemic IL-12 was abandoned years ago due to severe toxicity. Local mRNA delivery sidesteps that problem.

Multiple programs are in human trials. AstraZeneca's MEDI1191 showed partial responses in melanoma patients, including those resistant to existing immunotherapy. A Yale preclinical study found a single injection cured 60%+ of mice with melanoma.

There's also a newer approach: molecular on/off switches built into the mRNA itself. Researchers at Mount Sinai (published Nov 2025) engineered mRNA that detects cancer-specific microRNAs and only activates inside tumor cells. In mice: 100-fold higher expression in tumors, 380-fold lower in healthy organs. But this is still preclinical.

The catch: the on/off switch tech is mouse-only so far. The human trials rely on physically injecting into the tumor for selectivity, not molecular switches. Those are two different approaches at different stages, often blended together in popular coverage.

Read more:
https://www.mountsinai.org/about/newsroom/2025/scientists-develop-a-smarter-mrna-therapy-that-knows-which-cells-to-target
https://link.springer.com/article/10.1007/s00262-025-04105-0
https://doi.org/10.1158/1538-7445.am2023-ct004

#mRNA #cancer #immunotherapy

Just happened to notice I merged pull request #1000 for the Micro.blog backend today. About half of those are actually in the last year. I started using PRs way more often when AI could review them for me.

💊 A viral video calls benfotiamine (fat-soluble vitamin B1) a "revolutionary" anti-inflammatory breakthrough. The real story is more complicated.

The video references a real June 2026 study where researchers used computational drug repurposing to identify benfotiamine as a P2Y14R antagonist. P2Y14R is a genuine inflammatory target. The study tested it in mice with colitis and gouty arthritis and reported positive results.

But here's what the video leaves out:

1. The study is preclinical (mice + computer models), with 0 citations. "Revolutionary" is a stretch.

2. Multiple human RCTs found NO anti-inflammatory effect. A 24-month trial (300mg/day) in Type 1 diabetics showed no change in inflammatory markers. A 12-week trial (900mg/day) in diabetic nephropathy found no effect on inflammation. The BOND study (12 months, 600mg/day) found no effect on neuropathy outcomes.

3. The creator, Dr. Rick Pescatore, founded BellyMD, a supplement company selling benfotiamine products. This financial relationship is not disclosed in the video.

Benfotiamine is real, safe, and inexpensive. The preclinical anti-inflammatory science is genuine. But human trials have not confirmed the anti-inflammatory claims, and the "breakthrough" framing serves a product the creator sells.

Read more:
https://pmc.ncbi.nlm.nih.gov/articles/PMC3329837/
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0040427
https://doi.org/10.1136/bmjdrc-2025-005773

#science #medicine #supplements

🔬 Astronomers have been using machine learning since the late 1980s, decades before ChatGPT existed. The earliest neural networks in astronomy classified stars vs galaxies on photographic plates.

Here's what's interesting about how AI actually works in astronomy:

The AI used in astronomy is almost never a chatbot. It's specialized ML models trained for narrow tasks like classifying galaxies by shape or estimating parameters of gravitational wave signals. These models train on hundreds of thousands of already-classified galaxies, learn which features cluster together, then categorize new ones with a confidence score.

A huge application is outlier detection. The Vera C. Rubin Observatory will generate about 7 million alerts per night (tens of billions over its 10-year survey). Most are ordinary or noise. ML systems rank alerts so astronomers can focus on the genuinely unusual ones.

And no, AI isn't replacing astronomers. There are roughly 10,000 professional astronomers worldwide. The volume of data already far exceeds what humans could ever manually inspect.

Read more:
https://rubinobservatory.org/explore/how-rubin-works/alerts
https://pmc.ncbi.nlm.nih.gov/articles/PMC10230190/
https://ar5iv.labs.arxiv.org/html/1912.02934

#astronomy #MachineLearning #science

@michaelgemar Really good.

Watched: The Odyssey. Didn’t end up seeing it in IMAX, but hope to eventually. 🍿

I'll be doing some streaming over on my gaming channel if you're interested in hanging out. https://youtube.com/live/ORZW2moV6cI?feature=share

🌊 In September 2023, seismometers around the world detected a mysterious signal repeating every 92 seconds for nine straight days. The source: a standing wave trapped inside a Greenland fjord.

Here is what happened. A glacier in East Greenland had been thinning for years, no longer supporting the mountain slope above it. On September 16, about 25 million cubic meters of rock and ice collapsed into Dickson Fjord, triggering a tsunami with a runup height of 200 meters near the impact.

The fjord is 540 meters deep, 2.7 km wide, with walls taller than 1,800 meters. That geometry trapped the wave energy. The tsunami stabilized into a seiche, a standing wave sloshing back and forth across the fjord every 90 seconds. Each oscillation shifted millions of tons of water, loading and unloading the crust below, generating seismic waves detected globally.

The signal decayed extraordinarily slowly. It was still detectable after 9 days. A second, smaller event happened October 11 from another landslide in the same gully.

In 2025, the SWOT satellite directly observed the seiche, confirming the theory. The study published in Science calls it the first evidence of fluid sloshing persisting for days without an external driver.

The landslide was preconditioned by climate-driven glacial thinning, the first such event recorded in East Greenland.

Read more:
https://www.science.org/doi/10.1126/science.adm9247
https://www.nature.com/articles/s41467-025-59851-7
https://www.nasa.gov/missions/swot/international-swot-satellite-spots-planet-rumbling-greenland-tsunami/

#geology #seismology #climatechange

Install time is mostly gated by NVME speed, USB key read speed, and single-thread performance.