Researchers aimed to design a therapeutic that merged the drug delivery abilities of microrobots with the biological hunting abilities of natural cells. They genetically engineered living human embryonic kidney cells to express a special molecule called tumor necrosis factor–related apoptosis-inducing ligand, or TRAIL. When TRAIL binds to so-called “death receptors” in cancer cells, it sparks a series of signals that cause the cell to undergo programmed death. Healthy cells, which have comparatively lower levels of death receptors, get left unharmed.

The researchers then outfitted the TRAIL-modified cells with tiny magnetic beads that enabled the team to magnetically navigate the cells to their targets. Across all kinds of cancer cells the team tested, including colon, brain, kidney, and ovarian cancer cells, the cell-based bot significantly harmed cancerous cells while leaving normal cells alive.

The authors envision the cell-based robot approach working for diseases beyond cancer. “Because of its versatility, the platform can be adapted to a wide range of biomedical applications while maintaining a high degree of specificity and targeting,” they explained.

In 1695, the Spanish conquered this, too, and forcibly relocated its people a few decades later. While descriptions of the city can be found in historical documents, its location—along with archaeologists’ ability to understand life in a stronghold of Maya resistance that endured for over a century—has been lost.

In 2023, researchers visited a site called Sol y Paraíso and found many small mounds that could have been the remains of houses, ceramics that fit the style made by the Maya during the late precolonial and early colonial period, and two natural springs, which matched Spanish descriptions of the area around the town.

Could this have been Sac Balam? New evidence presented last week at the Society for American Archaeology annual meeting in San Francisco bolsters the case. Archaeologist Yuko Shiratori revealed that she and her team had found ceramic fragments as well as a monkey figurine that most likely date to the same period as Sac Balam . Crucially, she also found an imposing stone wall, 16 meters long and 1 meter high, that matches Spanish accounts of the size of Sac Balam’s three communal buildings. “I wasn’t sure [the site was Sac Balam] until I found that wall,” she said.

More evidence will be needed to fully convince the archaeological community, but those who attended Shiratori’s talk say it’s a good start. She and colleagues will be heading back to Sol y Paraíso this summer to dig for more clues.

In a new paper published in Science, researchers describe a machine learning method that learns to recognize and remove noise by looking across many exposures at once . The approach, called ASTERIS, takes advantage of the fact that real astronomical signals stay consistent from one exposure to the next, while noise varies. By combining spatial information within each image with temporal information across exposures, the system can tease apart the two. It is trained in a self-supervised way, meaning it does not rely on ideal “clean” reference images, but instead learns directly from the data itself, focusing on the faintest, noisiest regions.

ASTERIS is “unlocking faint sources in terabytes of existing data without additional telescope time,” said co-author Zheng Cai in an interview with Sky & Telescope. When applied to data from the James Webb Space Telescope and the Subaru Telescope, the method pushed detection limits more than one magnitude deeper, allowing astronomers to reliably identify objects about 2.5 times fainter than before. In one test, the algorithm uncovered roughly three times as many candidate galaxies at extreme distances compared with conventional methods.

For astronomers trying to map the early universe, that extra depth could translate into a clearer picture of how the first galaxies formed and evolved.

Hack’s law is named after observations by U.S. Geological Survey researcher John T. Hack in 1957. He noticed that a number of river branches, or tributaries, in the eastern United States could be described by

L = 1.4 A^0.6

which is a short way of saying that the longest length (L) tributary is scaled to its drainage area (A). The equation was applied to rivers in other places, and it kept working, aside from the exponent being sometimes a little below or above 0.6, depending on whether drainage growth was more side-to-side or elongated. Being able to describe complex river networks in this simple way made it much easier to estimate things like seasonal flux, landscape change, and flood risk.

The study by Tian Dong and colleagues flips this idea around. If an equation can describe the tributary branches at a river’s head, shouldn’t one also describe the distributary branches at its mouth? The patterns look very similar after all. This is not a new question. But answers for deltas have been elusive, probably because of one key difference between the head of a river and its mouth: topographic relief. Very low relief across deltas makes it hard to see and measure the channels themselves, not to mention that many of these channels are submerged. Early work has therefore focused on smaller-scale deltas and laboratory or numerical simulations.

The researchers were able to take a global approach by applying graph theory to satellite-imaged channel patterns across 30 river deltas. They systematically tagged “nodes” where channels split and “edges” at channel boundaries, which allowed the authors to extract measurements for length and area across nearly 6000 points. Plotted together, their data form a trend defined by this equation:

L = 1.43 A^0.60

which, nearly 70 years later, looks a lot like Hack’s law!

What is surprising is that the constant (1.4) and exponent (0.6) are so similar, despite tributaries and distributaries being built by fundamentally different systems. Tributary systems are convergent, where flow comes together and accelerates; distributary systems are divergent, where flow splits and decelerates. Specifically, the A for distributary systems is related to area of nourishment (deposition), not drainage and erosion.

Like Hack’s law for rivers, the equation for deltas changes slightly when the data are broken into certain subsets, an indication that local processes can help control the shape of a delta’s nourishment areas. Overall, as for rivers, the equations presented in this paper provide a framework for understanding how deltas organize over time to build and change their landscapes.

For me, this study was a reminder there are still fundamental discoveries to be made in areas we’ve long studied, where time and early tests and certain tools can come together to launch an old idea in new directions.

With contributions from Hannah Richter, Michael Price, Ana Georgescu, and Jocelyn Kaiser

Do you have a burning science question you can’t seem to find a good answer for? Submit it to Ask Science! Selected questions will receive responses from Science editors right here in ScienceAdviser.

Want more? Catch up on past issues of ScienceAdviser.

If you were forwarded this newsletter by a friend, you can subscribe for free here.