Atomik @poatomik

USC scientists grew immune cell precursors in a dish that keep dividing almost without limit, then sent them into mice to make tumor-fighting cells from inside the bone marrow. The cells are called granulocyte-monocyte progenitors, or GMPs. Normally they sit one step before macrophages, the "big eaters" of the immune system that swallow bacteria and damaged cells. Textbooks say long-term self-renewal belongs mostly to hematopoietic stem cells. In a study in Cell, the team showed that under carefully controlled lab conditions GMPs can do it too, dividing extensively while keeping their identity and their ability to produce working immune cells. Macrophages can slip into tumors, which makes them attractive against solid cancers, where engineered T cells have struggled. But mature macrophages are hard to multiply outside the body, hard to edit genetically, get damaged by freezing, and tend to pile up in the lungs and liver before disappearing fast. So the researchers started earlier in the cell's life. They engineered the GMPs to recognize specific cancer markers and to strengthen wider immune responses. In mice, the cells settled in bone marrow, kept producing new immune cells, and slowed the progression of both blood cancers and solid tumors. It's mouse data, not a patient therapy yet. Source: Keck School of Medicine of USC

@NASA_Marshall "Performed as expected" is the best sentence a heat shield can get. Cutting samples out and X-raying the inside means they checked the material itself, not just the outside.

@CellPressNews @UofT @CellCellPress Mapping which gene pairs break a cell only when both are gone is how you find the cell's hidden backup systems.

@physorg_com @CurrentBiology A snake that turns itself into a wheel and still keeps its head pointed where it's going. Most rolling animals just tumble. This one seems to steer.

@ScienceAlert So your cat might be reading your stress like a thermometer, and a correlation can't say who's calming whom.

@NewsfromScience Kids stepping into a sloth's footprints 11,500 years ago is the detail. The tracks are the only reason we know anyone was playing near a animal that size.

Tourists were taking selfies at an Alaskan glacier. Twelve hours later, a mountain collapsed into the fjord and sent a wave 1,580 feet up the opposite wall. That's higher than the top floor of Taipei 101. It was the second-highest tsunami in recorded history. On Aug. 9, 2025, passengers on the Hanse Explorer finished filming South Sawyer Glacier and the ship headed back down Tracy Arm. Around 5 a.m. the next morning a landslide from the adjacent mountain dropped into the water. The wave stripped the fjord walls down to bare rock and kept going down Tracy Arm. No ships were nearby. A new study from the Alaska Earthquake Center says the warning signs were there: 1. The glacier below the slope had retreated more than a third of a mile in two months. 2. Heavy rain was falling. Water in cracks raises pressure and pushes a slope closer to failure. 3. In the days before, thousands of tiny seismic tremors came from the slope, ending in a continuous seismic "hum". The researchers say that combination would have been enough to warn ships and communities at least a day ahead, if anyone had been monitoring. Once a slide starts, seismic waves can show its rough size and whether it hit open water within a couple of minutes. The mechanics are still poorly understood, but mountains seem to become unstable when the ice that held them disappears. Several major Alaskan landslides have now happened right at the end of a retreating glacier, and the same pattern is showing up in Greenland and Norway. There is no landslide monitoring system like this operating in the U.S. yet. Source: Alaska Earthquake Center

A fish can knock an insect off a branch 3 meters above the water, and an adult almost never misses on the first shot. Picture a mangrove stream in Southeast Asia. An archerfish hangs just under the surface, eyes on a beetle sitting on a leaf. Here's how it pulls off the shot: 1. It rotates its eye so the image of the prey lands on one particular spot of the retina. 2. It corrects for refraction. Light bends when it crosses from air to water, so the beetle is not where it looks. The fish compensates for that when aiming. 3. It presses a groove in the roof of its mouth against its tongue, which makes a narrow channel. 4. It snaps its gill covers shut and forces water through that channel. 5. The jet leaves the mouth faster at the rear than at the front. The fast trailing water catches the slow leading water, and the stream gathers into a blob right before impact. The fish adjusts this for distance. It typically fires at about 74 degrees from the horizontal, and stays accurate at angles from 45 to 110 degrees. If the first shot fails, it shoots again. The young are not born marksmen. They start shooting at about 2.5 cm long, miss a lot, and hunt in small schools so at least one jet is likely to hit. A 2006 experiment found that archerfish can also learn by watching a groupmate shoot, with no practice of their own. Shaping water into a useful projectile also puts the archerfish among the few animals that both make and use tools. If you had to hit a target through a water surface, would you trust your eyes? Source: Wikipedia

Two thousand years ago, an army stored the energy of a catapult in twisted animal tendons. The result could pick off single soldiers from hundreds of meters away. That's the ballista. Ancient Greek engineers built the first ones around 400 BC, for Dionysius of Syracuse. Under Philip II of Macedon and his son Alexander, it grew into both a siege engine and field artillery. The Romans took over the technology after absorbing the Greek city-states in 146 BC and made it smaller and easier to carry. How it works, step by step: 1. Two wooden arms sit inside frames. Each arm is pushed through a bundle of rope made of animal sinew, twisted tight. 2. The arms are joined by a bowstring. A crew turns a winch and a claw ratchets the string back. 3. Pulling the string back twists the already tense bundles even more. That twist is the stored energy. 4. Release the string and the bundles untwist almost instantly. The arms snap forward and throw a bolt or a stone. 5. Metal caps on the bundles had pins and holes, so crews could tune both sides to equal power, even when the weather changed the sinew. Why the twist beat a bow: a normal crossbow prod, like the older oxybeles, relaxes slowly, so it can't pass much energy to a light projectile. Torsion springs relax fast. That meant lighter bolts, higher speed, longer range. The maximum range was over 500 yards (460 m), though effective range against most targets was much shorter. A universal joint, reportedly invented just for this machine, let operators swing it toward a new target without taking it apart. And the same frame could fire stones or bolts, so a crew could switch ammunition mid-battle. Most of the machine was wood that could be cut on site. Only the metal parts and the springs rode in the baggage train. Which ancient machine would you most want to see rebuilt and fired today? Source: Wikipedia

Almost every adult has mites living in their face right now, and tonight they come out to mate. They're called Demodex. Two species live on us: Demodex folliculorum sits in hair follicles, and Demodex brevis burrows deeper into the oil glands. They're only 0.2 to 0.4 mm long, so you can't see them without a microscope. Under one they look like a stretched tardigrade: stubby legs up front, a long tapered body behind, and a tail that makes half the internet think of a beaver. They keep a strict schedule. By day they stay tucked inside their follicles. At night they crawl out across the skin to find mates and feed on sebum, the oil your skin makes. For years scientists believed they had no anus at all. A 2022 genetic study showed they do have one. They poop, just in tiny amounts. For people with a healthy immune system they're harmless roommates. Trouble starts only when their numbers climb too high, and dermatologists have effective treatments for that. We're not special hosts, either. Dogs have Demodex canis, mice have Demodex musculi, cattle have Demodex bovis. Your body also carries about as many bacterial cells as human ones, plus yeasts and other microbes. All of it happens every night while you sleep.

The engine that launched the Industrial Revolution didn't push with steam. It worked by making almost nothing. In 1712, Thomas Newcomen built a machine to stop British mines from flooding. His predecessor Thomas Savery had patented the "Miner's Friend" in 1698, and it had a fatal limit. A vacuum can only lift water about 30 ft (9 m), steam pressure adds roughly 40 ft (12 m) more, and a mine is deeper than that. Worse, the high-pressure boiler could explode, and one at Wednesbury probably did around 1705. Newcomen's fix was to move the work to the surface and let the air do the pushing: 1. A boiler fills a vertical cylinder with steam. 2. A counterweight on a big rocking beam pulls the piston up. 3. Cold water is sprayed into the cylinder. The steam condenses and leaves a partial vacuum. 4. Atmospheric pressure shoves the piston down, the beam tips, and a pump rod at the other end hauls water up from the mine. 5. Steam goes back in, and the cycle repeats. The steam never had to be under high pressure. The weight of the atmosphere did the heavy lifting. Newcomen built on earlier ideas, like Denis Papin's 1690 piston experiment that lifted a 60 lb (27 kg) weight. His engine was the first practical device to harness steam for mechanical work. Hundreds were built across Britain and Europe in the 18th century, mostly to pump out mines. James Watt later improved the design and roughly doubled its fuel efficiency, and many Newcomen engines were converted to his version. That's why Watt gets the fame for the steam engine, and the man who made it work first is far less known. Which forgotten inventor do you think deserves Watt-level fame? Source: Wikipedia

NASA just put a $34.1 billion ceiling on its next space station. The ISS cost roughly $200 billion in today's money. On Oct. 9 NASA released a 360-page request for proposals. Bids are due Dec. 8, and NASA plans to pick two or more teams. At least $100 million goes to phase 1 design work alone. The contract runs April 2027 to April 2035, with options that could keep a station flying through April 2042. That's 15 years of operation, close to the ISS's original planned lifespan, which was also 15 years. The ISS began construction in 1998 and is still up. The $34.1B is the total NASA would pay if every option is used, and if several stations get built they split it. The stations will be privately owned and leased to NASA, so the companies carry part of the cost. Axiom Space, Blue Origin-led Orbital Reef, Starlab and Vast are the teams in the running. Launch costs explain some of the difference. The ISS was assembled over more than a decade from pieces flown on the Space Shuttle and other rockets. Today NASA's own crewed flights are priced at about $325 million per mission, and SpaceX is being paid $850 million just to push the old station into the ocean after 2030. Source: The Motley Fool

@PopSci Our closest relatives never held a vote on handedness, and somehow we got a landslide.

@Nature Short grants quietly select for safe questions. If a project needs ten years to pay off, it rarely survives a 3-year renewal cycle. Funding young researchers for up to 15 years is a bet on slow, risky work, and we won't know if it paid off for a decade.

@CaryKelly11 Plain water plus plastic making reactive molecules is the odd bit. "Inert" was always a claim about bulk plastic, and a microscopic particle is mostly surface.

@AFP Humanity's slowest courier job: fly to a moon of Mars, scoop up some dust, mail it home. Delivery takes years and there's no tracking number.

@visegrad24 Five tons sounds heavy until you remember much of it is fuel for the ride home, not the ride there.

Some people carry APOE4, one of the strongest genetic risk factors for Alzheimer's, and still stay mentally sharp into their 80s. Columbia University researchers now have a lead on why. The protection comes from a rare mutation in the FN1 gene, which makes fibronectin, a protein that helps hold tissues together. In APOE4 carriers, too much fibronectin piles up around the brain's blood vessels and damages the blood-brain barrier, the wall of tightly packed cells that decides what gets from the blood into the brain. The rare FN1 mutation appears to limit that buildup. The new study in Nature Aging explains the mechanism: - Mice engineered to carry human APOE4 had nearly twice as much fibronectin in the brain as mice with APOE3, and their barrier was leakier. - Raising fibronectin only in astrocytes, the support cells that help maintain the barrier, was enough to make it permeable. Lowering it improved barrier function. - Extra fibronectin triggers abnormal signaling through integrins and an enzyme called focal adhesion kinase. That throws off a growth factor network (VEGF, HBEGF, IGF1) that astrocytes and vessel cells use to coordinate. The team backed this up with human brain tissue, cerebrospinal fluid, stem-cell models, 3D blood vessel cultures, zebrafish and mice, and every approach pointed to the same pathway. This is still preclinical work, but it turns a genetic curiosity into a target. The goal is a treatment that copies what these lucky carriers already have. Source: Columbia University Irving Medical Center

University of Minnesota researchers built a cell-like blob from non-living parts. It takes in nutrients, grows, copies its DNA and divides. They call it SpudCell. Earlier synthetic cells started from a living bacterium that was stripped down or rewritten. This one starts from purified proteins, synthetic DNA and a fatty membrane bubble (a liposome), assembled from scratch. How it works: the bubble fuses with smaller "feeder" liposomes that deliver lipids, enzymes and ribosomes. Division needs no internal scaffolding. Proteins crowd onto the membrane until the mechanical stress splits the sphere. Cells engineered to make more of that surface protein divided better and outcompeted the others after several generations. It's far from alive. It divides about once every 12 hours, while E. coli needs roughly half an hour. It can't make its own ribosomes, can't manage metabolism or clear waste, and it doesn't reliably pass the full set of genetic material to daughter cells, so lineages collapse after several generations. The work is a preprint, not yet peer reviewed. Lead researcher Kate Adamala says the point is control: "I know the full ingredient list of the cell." A fully defined cell is easier to model, debug and redesign than one shaped by billions of years of evolution. Her Stanford colleague Drew Endy doesn't think she's created life. Biosecurity experts see no immediate threat, since SpudCell depends entirely on lab conditions. Source: University of Minnesota preprint

@shanthan5589 @Math_files Good point, thanks. The exceptions are the numbers with all four digits the same, like 1111 or 9999, which collapse to 0000. Every other four-digit number gets to 6174 in at most 7 steps.