Mitochondrial complex I (also known as NADH:ubiquinone oxidoreductase) contributes to cellular energy production by transferring electrons from NADH to ubiquinone coupled to proton translocation across the membrane1,2. It is the largest protein assembly of the respiratory chain with a total mass of 970 kilodaltons3. Scientists now present a nearly complete atomic structure of ovine (Ovis aries) mitochondrial complex I at 3.9 Å resolution, solved by cryo-electron microscopy with cross-linking and mass-spectrometry mapping experiments. All 14 conserved core subunits and 31 mitochondria-specific supernumerary subunits are resolved within the L-shaped molecule. The hydrophilic matrix arm comprises flavin mononucleotide and 8 iron–sulfur clusters involved in electron transfer, and the membrane arm contains 78 transmembrane helices, mostly contributed by antiporter-like subunits involved in proton translocation.
Supernumerary subunits form an interlinked, stabilizing shell around the conserved core. Tightly bound lipids (including cardiolipins) further stabilize interactions between the hydrophobic subunits. Subunits with possible regulatory roles contain additional cofactors, NADPH and two phosphopantetheine molecules, which are shown to be involved in inter-subunit interactions. The researchers observe two different conformations of the complex, which may be related to the conformationally driven coupling mechanism and to the active–deactive transition of the enzyme.
The obtained structure of the complex provides insight into the mechanism, assembly, maturation and dysfunction of mitochondrial complex I, and allows detailed molecular analysis of disease-causing mutations.
What is a leech lattice? It turns out that it’s the only even integral unimodular lattice with no roots in fewer than 32 dimensions. Each of these terms will take some explaining:
A lattice, as mentioned at the beginning of this article, is a set of vectors in Euclidean space which is closed under addition.
A lattice is integral if the dot product of any two vectors is an integer.
An integral lattice is described as even if the squared norm of any vectors is an even integer. Otherwise, it is an odd integral lattice.
A lattice is unimodular if the fundamental parallelotope has unit volume, or the determinant of the generating matrix is ±1. Equivalently, the lattice has a density of one point per unit volume. Even integral unimodular lattices only occur in dimensions divisible by 8, with the unique example in 8 dimensions being the E8 lattice.
For an even unimodular lattice, a root is a vector of squared norm 2. The Leech lattice has none of these, so the shortest nonzero vectors have a squared norm of 4 (or, equivalently, a length of 2).
Because the shortest distance between two lattice points is equal to 2, that means that we can place a ball of unit radius centred on each lattice point. Each sphere touches precisely 196560 others, which is the maximum for 24 dimensions. The kissing number problemhas also been solved in 1, 2, 3, 4 and 8 dimensions, with values of 2, 6, 12, 24 and 240, respectively, attained by the integer lattice, hexagonal lattice, FCC lattice, D4 lattice and E8 lattice, respectively.
If we enlarge the spheres to a radius of √2, they will overlap and cover all space. Establishing that the covering radius is equal to √2 is non-trivial, and involves analysing all 23 types of deep hole (points maximally distant from lattice points) encountered in the Leech lattice. Each type of deep hole corresponds to one of the 23 other even unimodular 24-dimensional lattices, known as Niemeier lattices.
Reconstructions of single cells highlight how far they can reach into the brain.
The 70 million neurons in the mouse brain look like a tangled mess, but researchers are beginning to unravel the individual threads that carry messages across the organ. A 3D brain map released on 27 October, called MouseLight, allows researchers to trace the paths of single neurons and could eventually reveal how the mind assembles information.
The map contains 300 neurons and researchers plan to add another 700 in the next year. “A thousand is just beginning to scratch the surface,” says Nelson Spruston, a neuroscientist at the Howard Hughes Medical Institute (HHMI) Janelia Research Campus in Ashburn, Virginia.
To create the maps, Spruston and HHMI neuroscientist Jayaram Chandrashekar injected mouse brains with viruses that infect only a few cells at a time, prompting them to produce fluorescent proteins1. The team made the organs transparent using a sugar-alcohol treatment to obtain an unobstructed view of the glowing neurons, and then scanned each brain with a high-resolution microscope. Computer programs created 3D models of the glowing neurons and their projections, called axons, which can be half a meterlong and branch like a tree.
MouseLight has already revealed new information, including the surprisingly extensive number of brain regions that a single axon can reach. For instance, four neurons associated with taste stretch into the region that controls movement and another area related to touch. Chandrashekar says the group is now working on identifying which genes each neuron expresses, which will help to pin down their function.
“This is a tremendous project,” says Hongkui Zeng, a molecular biologist at the Allen Institute for Brain Science in Seattle, Washington, who plans to collaborate with the Janelia group on MouseLight. The Janelia technique is similar to one that Zeng and her colleagues developed using a line of mice genetically engineered so that a certain drug activates glowing proteins in a handful of their neurons.
MouseLight is just one of several methods being used to reconstruct individual neurons, says Rafael Yuste, a neurobiologist at Columbia University in New York City. Accurately labelling neurons with markers such as fluorescence, he says, will probably be the key challenge in the eventual goal of creating a “census” of different cell types in the brain.
UCLA researchers have designed a device that can use solar energy to inexpensively and efficiently create and store energy, which could be used to power electronic devices, and to create hydrogen fuel for eco-friendly cars.
The device could make hydrogen cars affordable for many more consumers because it produces hydrogen using nickel, iron and cobalt — elements that are much more abundant and less expensive than the platinum and other precious metals that are currently used to produce hydrogen fuel.
“Hydrogen is a great fuel for vehicles: It is the cleanest fuel known, it’s cheap and it puts no pollutants into the air — just water,” said Richard Kaner, the study’s senior author and a UCLA distinguished professor of chemistry and biochemistry, and of materials science and engineering. “And this could dramatically lower the cost of hydrogen cars.”
The technology, described in a paper in the journal Energy Storage Materials, could be especially useful in rural areas, or to military units serving in remote locations. “People need fuel to run their vehicles and electricity to run their devices,” Kaner said. “Now you can make both electricity and fuel with a single device.”
It could also be part of a solution for large cities that need ways to store surplus electricity from their electrical grids. “If you could convert electricity to hydrogen, you could store it indefinitely,” said Kaner, who also is a member of UCLA’s California NanoSystems Institute.
Synthetic biologists are converting microbial cells into living devices that are able to perform useful tasks ranging from the production of drugs, fine chemicals and biofuels to detecting disease-causing agents and releasing therapeutic molecules inside the body. To accomplish this, they fit cells with artificial molecular machinery that can sense stimuli such as toxins in the environment, metabolite levels or inflammatory signals. Much like electronic circuits, these synthetic biological circuits can process information and make logic-guided decisions. Unlike their electronic counterparts, however, biological circuits must be fabricated from the molecular components that cells can produce, and they must operate in the crowded and ever-changing environment within each cell.
Similar to how computer scientists use logical language to have their programs make accurate AND, OR and NOT decisions towards a final goal, “Ribocomputing Devices” (stylized here in yellow) developed by a team at the Wyss Institute can now be used by synthetic biologists to sense and interpret multiple signals in cells and logically instruct their ribosomes (stylized in blue and green) to produce different proteins.
So far, synthetic biological circuits can only sense a handful of signals, giving them an incomplete picture of conditions in the host cell. They are also built out of several moving parts in the form of different types of molecules, such as DNAs, RNAs, and proteins, that must find, bind and work together to sense and process signals. Identifying molecules that cooperate well with one another is difficult and makes development of new biological circuits a time-consuming and often unpredictable process.
As reported in Nature, a team at Harvard’s Wyss Institute for Biologically Inspired Engineering is now presenting an all-in-one solution that imbues a molecule of ‘ribo’ nucleic acid or RNA with the capacity to sense multiple signals and make logical decisions to control protein production with high precision. The study’s approach resulted in a genetically encodable RNA nano-device that can perform an unprecedented 12-input logic operation to accurately regulate the expression of a fluorescent reporter protein in E. coli bacteria only when encountering a complex, user-prescribed profile of intra-cellular stimuli. Such programmable nano-devices may allow researchers to construct more sophisticated synthetic biological circuits, enabling them to analyze complex cellular environments efficiently and to respond accurately.
“We demonstrate that an RNA molecule can be engineered into a programmable and logically acting “Ribocomputing Device,” said Wyss Institute Core Faculty member Peng Yin, Ph.D., who led the study and is also Professor of Systems Biology at Harvard Medical School. “This breakthrough at the interface of nanotechnology and synthetic biology will enable us to design more reliable synthetic biological circuits that are much more conscious of the influences in their environment relevant to specific goals.”
A new method has been developed that could help scientists identify the “lost 99 per cent” of microplastics in the world’s oceans. Microplastics are fragments of plastic measuring less than 5 millimeters. They mainly arise from large plastic objects, which accumulate in the oceans and slowly break down into smaller pieces. These tiny plastics are by far the most common type of marine pollution, but their size means most of them aren’t being taken into account when scientists attempt to analyze their prevalence.
One study estimated that 99 per cent of plastic in the ocean remains unaccounted for. Now, scientists have used a fluorescent dye that specifically binds to plastic in order to accurately count the very smallest plastic fragments. They found many more microplastics than were previously estimated.
“Using this method, a huge series of samples can be viewed and analyzed very quickly, to obtain large amounts of data on the quantities of small microplastics in seawater or, effectively, in any environmental sample,” said University of Warwick researcher Gabriel Erni-Cassola, lead author of the Environmental Science & Technology study.
A country’s ability to improve its standard of living over time depends almost entirely on its ability to raise its output per worker. That’s why Nobel laureate Paul Krugman concluded that productivity isn’t everything – but in the long run it is almost everything. Instead of wasting the nation’s time focusing on the non-existent threat of the deficit, the chancellor, Philip Hammond, conceded the “everything” that Mr Krugman had identified: British productivity has stalled and as a result workers’ real wages will be lower than when the recession began. Before the crash, we would have expected living standards to double every 40 years. Now that will take 80. That means lost decades for millions of ordinary people.
It’s easy to become overly pessimistic. The epoch of enormous economic progress that characterized the 20th century is not over; we are suffering from seven years of government failure where ministers thought their job was to watch the economy and suffer passively from capitalism’s inevitable cycles. Rather than take a view of the economy and by fiscal action seek to secure prosperity for all, ministers embarked on a highly ideological agenda of dismantling the state and the protections afforded to workers – arguing erroneously that these were holding back the state. Mr Hammond has been forced to alter course because his party’s reckless policies had jeopardized the long-term improvement in the national standard of life.
Britain’s problem is that its most productive regions are also the most unequal. Workers in London and many towns in the south-east produce more each hour than those in hyper-productive Germany. However, those cities recorded the highest levels of inequality. Who in their right mind would think it would be a good idea to spread around the country a model built on baristas and barristers? What all politicians agree on is that to raise productivity Britain will need to embrace technological advance. After big leaps in artificial intelligence, as our magazine cover story charts, robots have finally reached a maturity that allows mass deployment.
Today’s wave of automation is different in the sense it is “intelligent”. Such cleverness alters the effect on labor. Consider the introduction of driverless vehicles, which Mr Hammond said would mean a million people who drive for a living having to find new jobs. A car and a driver have to be combined to work as a taxi. A driver with two taxis creates no extra output. A faster car or a satnav doesn’t change much. But if a passenger can ride in a driverless car, then the owners of taxi firms will invest in technology rather than drivers. A few winner-take-all entrepreneurs ought not to be able to monopolize all the profits of technology while workers’ wages stagnate. There’s a good case for a universal basic income for humans, if vassal robots take all the jobs.
Humans tend to live in hierarchies, which rely on good leadership that ideally benefits the collective whole. There is little doubt that such systems are subject to frequent flaws, given that leaders may be vulnerable, corrupt, or self-interested. Even the most popular leaders have their detractors.
But most species of schooling fish and flocking birds do not have such problems. New research published in the journal PLOS Computational Biology finds that schooling fish switch their attention from neighbor to neighbor for seamless collective movement. The findings are helping scientists to design leaderless, self-organizing systems, such as swarms of drones.
The study focused on a common aquarium fish known as the rummy-nose tetra. When schooling, they can almost look like a single organism, given their coordinated movements. Each individual fish, however, has its own unique personality that can lead to relationship drama.
"While the movement of the fish gives the impression of the school as being very similar to a single organism, there are nonetheless conflicts within these groups: Rummy-nose tetras will still fight over food, or when mating," co-author Andrea Perna of the University of Roehampton in London told Seeker.
Despite the problems, fish do not drop out of 'school'. "Schooling is advantageous to each individual fish because they are less likely to be eaten by predators when they are in a school and because they can get information about food and about the environment from their neighbors,” she continued. “This is probably the main reason why they each individually decide to keep staying in the school."
A team of Department of Energy (DOE) scientists at the Joint Center for Energy Storage Research (JCESR) has discovered the fastest magnesium-ion solid-state conductor, a major step towards making solid-state magnesium-ion batteries that are both energy dense and safe.
The electrolyte, which carries charge back and forth between the battery's cathode and anode, is a liquid in all commercial batteries, which makes them potentially flammable, especially in lithium-ion batteries. A solid-state conductor, which has the potential to become an electrolyte, would be far more fire-resistant.
Researchers at DOE's Lawrence Berkeley National Laboratory (Berkeley Lab) and Argonne National Laboratory were working on a magnesium battery, which offers higher energy density than lithium, but were stymied by the dearth of good options for a liquid electrolyte, most of which tend to be corrosive against other parts of the battery. "Magnesium is such a new technology, it doesn't have any good liquid electrolytes," said Gerbrand Ceder, a Berkeley Lab Senior Faculty Scientist. "We thought, why not leapfrog and make a solid-state electrolyte?"
The material they came up with, magnesium scandium selenide spinel, has magnesium mobility comparable to solid-state electrolytes for lithium batteries. Their findings were reported in Nature Communications in a paper titled, "High magnesium mobility in ternary spinel chalcogenides." JCESR, a DOE Innovation Hub, sponsored the study, and the lead authors are Pieremanuele Canepa and Shou-Hang Bo, postdoctoral fellows at Berkeley Lab.
"With the help of a concerted effort bringing together computational materials science methodologies, synthesis, and a variety of characterization techniques, we have identified a new class of solid conductors that can transport magnesium ions at unprecedented speed," Canepa said.
’Omics bashing is in fashion. In the past year, The New York Times and The Wall Street Journalhave run pieces poking fun at the proliferation of scientific words ending in -ome, which now number in the thousands. One scientist has created a badomics generator, which randomly adds the suffix to a list of biological terms and generates eerily plausible titles for scientific papers (example: ‘Sequencing the bacteriostaticome reveals insights into evolution and the environment’). Jonathan Eisen, a microbiologist at the University of California, Davis, regularly announces awards for unnecessary additions to the scientific vocabulary on his blog (recent winner: CircadiOmics, for genes involved in daily circadian rhythms).
Botanist Hans Winkler had no idea what he was starting back in 1920, when he proposed the term ‘genome’ to refer to a set of chromosomes. Other ’omes existed even then, such as biome (collection of living things) and rhizome (system of roots), many of them based on the Greek suffix ‘-ome’ — meaning, roughly, ‘having the nature of’. But it was the glamorization of ‘genome’ by megabuck initiatives such as the Human Genome Project that really set the trend in motion, says Alexa McCray, a linguist and medical informatician at Harvard Medical School in Boston, Massachusetts. “By virtue of that suffix, you are saying that you are part of a brand new exciting science.”
Researchers also recognize the marketing potential of an inspirational syllable, says Eisen. “People are saying that it’s its own field and that it deserves its own funding agency,” he says. But although some ’omes raise an eyebrow — museomics (sequencing projects on archived samples) and the tongue-in-cheek ciliomics (study of the wriggling hairlike projections on some cells) — scientists insist that at least some ’omes serve a good purpose. “Most of them will not make sense and some will make sense, so a balance should be in place,” says Eugene Kolker, chief data officer at Seattle Children’s Hospital in Washington, and founding editor of the journal Omics. “If we just laugh about different new terms, that’s not good.”
Ideally, branding an area as an ’ome helps to encourage big ideas, define research questions and inspire analytical approaches to tackle them (see ‘Hot or not’). “I think -ome is a very important suffix. It’s the clarion call of genomics,” says Mark Gerstein, a computational biologist at Yale University in New Haven, Connecticut. “It’s the idea of everything, it’s the thing we find inspiring.” Here, Nature takes a look at five up-and-coming ’omes that represent new vistas in science.
A tour through the most studied genes in biology reveals some surprises.
Peter Kerpedjiev needed a crash course in genetics. A software engineer with some training in bioinformatics, he was pursuing a PhD and thought it would really help to know some fundamentals of biology. “If I wanted to have an intelligent conversation with someone, what genes do I need to know about?” he wondered.
Kerpedjiev went straight to the data. For years, the US National Library of Medicine (NLM) has been systematically tagging almost every paper in its popular PubMed database that contains some information about what a gene does. Kerpedjiev extracted all the papers marked as describing the structure, function or location of a gene or the protein it encodes.
Sorting through the records, he compiled a list of the most studied genes of all time — a sort of ‘top hits’ of the human genome, and several other genomes besides. Heading the list, he found, is a gene called TP53. Three years ago, when Kerpedjiev first did his analysis, researchers had scrutinized the gene or the protein it produces, p53, in some 6,600 papers. Today, that number is at about 8,500 and counting. On average, around two papers are published each day describing new details of the basic biology of TP53.
Its popularity shouldn’t come as news to most biologists. The gene is a tumor suppressor, and widely known as the ‘guardian of the genome’. It is mutated in roughly half of all human cancers. “That alone explains its staying power,” says Bert Vogelstein, a cancer geneticist at the Johns Hopkins University School of Medicine in Baltimore, Maryland. In cancer, he says, “there’s no gene more important than p53”.
But some chart-topping genes are less well known — including some that rose to prominence in bygone eras of genetic research, only to fall out of fashion as technology progressed. “The list was surprising,” says Kerpedjiev, now a postdoc studying genomic-data visualization at Harvard Medical School in Boston, Massachusetts. “Some genes were predictable; others were completely unexpected.”
In a recent study, published in Nature, ICFO researchers Nicolas Maring, Pau Farrera, Dr. Kutlu Kutluer, Dr. Margherita Mazzera, and Dr. Georg Heinze led by ICREA Prof. Hugues de Riedmatten, have achieved an elementary "hybrid" quantum network link and demonstrated for the first time photonic quantum communication between two very distinct quantum nodes placed in different laboratories, using a single photon as information carrier.
Today, quantum information networks are ramping up to become a disruptive technology that will provide radically new capabilities for information processing and communication. Recent research suggests that this quantum network revolution might be just around the corner.
The key elements of a Quantum Information Network are quantum nodes, that store and process the information, made up of matter systems like cold atomic gases or doped solids, among others, and communicating particles, mainly photons. While photons seem to be perfect information carriers, there is still uncertainty as to which matter system could be used as network node, as each system provides different functionalities. Therefore, the implementation of a hybrid network has been proposed, searching to combine the best capabilities of different material systems.
Past studies have documented reliable transfers of quantum information between identical nodes, but this is the first time this has ever been achieved with a "hybrid" network of nodes. The ICFO researchers have been able to come up with a solution to making a hybrid quantum network work and solve the challenge of a reliable transfer of quantum states between different quantum nodes via single photons. A single photon needs to interact strongly and in a noise-free environment with the heterogeneous nodes or matter systems, which generally function at different wavelengths and bandwidths. As Nicolas Maring states "it's like having nodes speaking in two different languages. In order for them to communicate, it is necessary to convert the single photon's properties so it can efficiently transfer all the information between these different nodes."
How did they solve the problem?
In their study, the ICFO researchers used two very distinct quantum nodes: the emitting node was a laser-cooled cloud of Rubidium atoms and the receiving node a crystal doped with Praseodymium ions. From the cold gas, they generated a quantum bit (qubit) encoded in a single photon with a very-narrow bandwidth and a wavelength of 780 nm. They then converted the photon to the telecommunication's wavelength of 1552 nm to demonstrate that this network could be completely compatible with the current telecom C-band range. Subsequently, they sent it through an optical fiber from one lab to the other. Once in the second lab, the photon's wavelength was converted to 606 nm in order to interact correctly and transfer the quantum state to the receiving doped crystal node. Upon interaction with the crystal, the photonic qubit was stored in the crystal for approximately 2.5 microseconds and retrieved with very high fidelity.
The results of the study have shown that two very different quantum systems can be connected and can communicate by means of a single photon. As ICREA Prof at ICFO Hugues de Riedmatten comments, "being able to connect quantum nodes with very different functionalities and capabilities and transmitting quantum bits by means of single photons between them represents an important milestone in the development of hybrid quantum networks." The ability to perform back- and forth-conversion of photonic qubits at the telecom C-band wavelength shows that these systems would be completely compatible with the current telecom networks.
Biohybrid bot could one day deliver drugs or do surgery.
For decades, engineers have been trying to build medical robots that can deliver drugs or do surgery inside the human body—a somewhat less fantastic version of the 1966 sci-fi film Fantastic Voyage. Now, scientists have manipulated spirulina, a microscopic plant and food supplement, to travel through people in response to magnetic signals. The biohybrid robot could one day carry drugs to specific parts of the body, minimizing side effects. What’s more, the robot—and its magnetic coat—appear to kill cancer cells.
Spirulina, an alga, looks like a tiny coiled spring at the microscopic level. Researchers had been trying, and succeeding to various degrees, to build bots out of rods, tubes, spheres, and even cages no bigger than a cell. Outfitting these tiny devices with an ample power supply has been quite a challenge, as most potential fuels are toxic to humans. Another problem is steering such a microrobot through the body’s maze of proteins and other molecules, which requires both a way to control its movements and to see where it is.
So Li Zhang, a materials scientist at the Chinese University of Hong Kong in Shatin, turned to magnetism—and living organisms. Magnetic fields created outside the body can penetrate living tissue without harm, allowing researchers to move magnetized objects around inside. For maximum mobility, a helical body propelled by twirling works best. Enter Spirulina. “It’s surprising that you can find in nature such a convenient structure and that it can behave so nicely,” says Peer Fischer, a physical chemist at the Max Planck Institute for Intelligent Systems in Stuttgart, Germany, who was not involved in the study.
Several years ago, Zhang and his colleagues used the alga as inspiration for a synthetic microbot, which worked to some degree. This time, the scientists decided to use the alga itself. They needed a way to track the robot in the body, and the alga produces a fluorescent glow. The researchers wondered whether they could follow the robot's course near the body surface by detecting this fluorescence, and then use a commonly used medical imaging technology called nuclear magnetic resonance (NMR) to track it in deeper parts of the body. NMR works by detecting magnetic particles given to a patient before the imaging takes place.
They developed a one-step method to magnetize the alga, coating millions of Spirulina with iron oxide nanoparticles. A longer dip time allows for more control, but a shorter dip time allows researchers to detect the fluorescence more readily. When the bot is too deep for that technique to work, NMR can still follow the robot’s course because of the coating, the researchers report today in Science Robotics. Using NMR, they observed the microrobots swarm in a rat’s stomach as directed by the magnetic field.
“It’s a step forward that you can track these swimmers in the body,” says Joseph Wang, a nanoengineer at the University of California, San Diego, who is developing a different sort of medical microbot. “And it’s biocompatible and low cost.”
This is Nicolas Loizeau's programmable computer implemented in Conway's game of life computing Fibonacci sequence. Soon artificial self-encoding life forms are possible.
DeepMind's Professor David Silver describes AlphaGo Zero, the latest evolution of AlphaGo, the first computer program to defeat a world champion at Go. There's been way too many fear-mongering news articles around the latest version of DeepMind's AlphaGo. Let's set the record straight, AlphaGo is an incredible technology and it's not terrifying at all. I'll go over the technical details of how AlphaGo really works; a mixture of deep learning and reinforcement learning.
Something amazing has happened. A couple of years ago, we closely followed the progress of AlphaGo, the distributed DeepMind algorithm which defeated Lee Sedol in four out of five games of the ancient board game Go. This has since been defeated by a variant,A lphaGo Zero, which is considerably stronger — after 20 days of training, it was able to win 100 out of 100 games against the version of AlphaGo which played against Sedol. After a further 20 days of training, it won 89 out of 100 games against a stronger instantiation of AlphaGo, namely the one which defeated the world champion Ke.
However, its superiority over the previous algorithm isn’t the most interesting aspect. What makes it interesting is that, unlike AlphaGo which both trained on human games and made use of hardcoded features (such as ‘liberties’), AlphaGo Zero is remarkably simple:
The algorithm has no external inputs, learning only from games against itself;
The input to the neural network is just 17 inputs, namely the parity of the turn, the indicator functions of white stones for the last 8 positions, and the indicator functions of black stones for the last 8 positions. (Storing the immediate history is a necessity due to ko rules.)
Instead of separate policy and value networks, the algorithm uses only one neural network;
Monte Carlo rollouts are ditched in favour of a feedback loop where the tree search evolves together with the network.
Read the Nature paper for more details. AlphaGo Zero was trained on just four tensor processing units (TPUs), which are fast hardware implementations of fixed-point limited-precision linear algebra. This is much more efficient (but less numerically precise) than a GPU, which is in turn much more efficient (but less flexible) than a CPU.
Is Simone Biles about to lose her job to a robot, too?
Here’s another thing robots can do better than you: backflips. Boston Dynamics, the MIT offshoot company now owned by Japanese tech giant SoftBank, showed off the latest iteration of their bipedal Atlas robot in a video released Thursday. At first viewing, the mobility of a 4 foot 9 inch, approximately 165-pound hydraulic machine is mind-boggling. But are robots going to steal Simone Biles’ job, too?
Atlas has made cheer-worthy progress since 2013, when Boston Dynamics debuted it at a robotics challenge sponsored by the Defense Advanced Research Projects Agency, and 2015, when it competed in the finals of the competition. Essentially, the federal government hosted a contest where humanoid robots had to complete a series of simple tasks useful in the case of a nuclear power plant disaster, like traveling up a one- to two-degree incline with scattered obstacles, shutting off a valve, opening a door to enter a building. It was a fail-fest. Making a basic bipedal robot is no small feat, but they were like high-tech toddlers. The competition’s finals spawned video reels of off-kilter robots crashing to the ground. By comparison, this version of Atlas is pretty impressive (and significantly less creepy than Boston Dynamics’ SpotMini, a spindly yellow robotic dog).
But we wanted to know how impressive, so Slate asked a gymnastics coach to rate Atlas’ parkour skills. According to gymnastics coach Aryan Mazloum, Atlas’ backflip—a back salto, if you want the technical term—is not bad. “It’s pretty fantastic to be able see a robot have the center of gravity and be able to not only just move, but literally flip and catch itself,” said Mazloum, a junior Olympic coach at Northern Virginia’s Capital Gymnastics National Training Center. He’s also working toward a Ph.D. in informatics at George Mason University.
The back salto, Mazloum explains, is “an intermediate skill” that coaches introduce in the fifth level of USA Gymnastics, when students tend to be 9 to 11 years old. For a robot, it takes incredible spatial awareness. In a back salto, says Mazloum, “you want to be able to go as high as you can, and you want to be able to land as close to where you take off as possible.” To do that, the gymnast has to squat, throw her arms up by her ears so her body is a straight line (in gymnast-speak, opening the shoulder angle and the hip), then contract into a “closed” position again. By these standards, Atlas’ trick is “not the cleanest flip,” explains Mazloum.
Here’s Mazloum’s critique: Atlas didn’t quite get to that open position, “so it didn’t really get the full vertical that we look for. That’s why it went backwards a little bit.” But, he adds, it’s “still astonishing that it did that, though.” By the way, at the end of the video, where Atlas falls? That’s again probably because it didn’t get enough height, which means it didn’t have the time to rotate, and then since the robot lacks toes, it couldn’t push into the ground to counterbalance.
Still, Mazloum gives the robot kudos: “It was a good landing, I’ll say that.” In gymnastics, you don’t score individual components, only full routines. But Mazloum made an exception for Atlas: 3.5/5 for its back salto.
Scientists at the University of Oxford have developed a new method to 3D-print laboratory-grown cells to form living structures.
The approach could revolutionize regenerative medicine, enabling the production of complex tissues and cartilage that would potentially support, repair, or augment diseased and damaged areas of the body.
Published in the journal Scientific Reports, an interdisciplinary team from the Department of Chemistry and the Department of Physiology, Anatomy, and Genetics at Oxford and the Centre for Molecular Medicine at Bristol, demonstrated how a range of human and animal cells can be printed into high-resolution tissue constructs.
Interest in 3D printing living tissues has grown in recent years, but, developing an effective way to use the technology has been difficult, particularly since accurately controlling the position of cells in 3D is hard to do. They often move within printed structures and the soft scaffolding printed to support the cells can collapse on itself. As a result, it remains a challenge to print high-resolution living tissues.
But, led by Hagan Bayley, professor of chemical biology in Oxford's Department of Chemistry, the team devised a way to produce tissues in self-contained cells that support the structures to keep their shape. The cells were contained within protective nanoliter droplets wrapped in a lipid coating that could be assembled, layer-by-layer, into living structures. Producing printed tissues in this way improves the survival rate of the individual cells, and allowed the team to improve on current techniques by building each tissue one drop at a time to a more favorable resolution.
To be useful, artificial tissues need to be able to mimic the behaviors and functions of the human body. The method enables the fabrication of patterned cellular constructs, which, once fully grown, mimic or potentially enhance natural tissues.
Dr. Alexander Graham, lead author and 3D bioprinting scientist at OxSyBio (Oxford Synthetic Biology), said: "We were aiming to fabricate three-dimensional living tissues that could display the basic behaviors and physiology found in natural organisms. To date, there are limited examples of printed tissues, which have the complex cellular architecture of native tissues. Hence, we focused on designing a high-resolution cell printing platform, from relatively inexpensive components, that could be used to reproducibly produce artificial tissues with appropriate complexity from a range of cells including stem cells."
Researchers at MIT have developed a synthetic gene circuit that triggers the body’s immune system to attack cancers when it detects signs of the disease. The circuit, which will only activate a therapeutic response when it detects two specific cancer markers, is described in a paper published today in the journal Cell.
Immunotherapy is widely seen as having considerable potential in the fight against a range of cancers. The approach has been demonstrated successfully in several recent clinical trials, according to Timothy Lu, associate professor of biological engineering and of electrical engineering and computer science at MIT and a member of MIT's Koch Institute for Integrative Cancer Research.
“There has been a lot of clinical data recently suggesting that if you can stimulate the immune system in the right way you can get it to recognize cancer,” says Lu, who is head of the Synthetic Biology Group in MIT’s Research Laboratory of Electronics. “Some of the best examples of this are what are called checkpoint inhibitors, where essentially cancers put up stop signs that prevent T-cells from killing them. There are antibodies that have been developed now that basically block those inhibitory signals and allow the immune system to act against the cancers.”
However, despite this success, the use of immunotherapy remains limited by the scarcity of tumor-specific antigens — substances that can trigger an immune system response to a particular type of cancer. The toxicity of some therapies, when delivered as a systemic treatment to the whole body, for example, is another obstacle.
What’s more, the treatments are not successful in all cases. Indeed, even in some of the most successful tests, only 30-40 percent of patients will respond to a given therapy, Lu says.
As a result, there is now a push to develop combination therapies, in which different but complementary treatments are used to boost the immune response. So, for example, if one type of immunotherapy is used to knock out an inhibitory signal produced by a cancer, and the tumor responds by up-regulating a second signal, an additional therapy could then be used to target this one as well, Lu says.
“Our belief is that there is a need to develop much more specific, targeted immunotherapies that work locally at the tumor site, rather than trying to treat the entire body systemically,” he says. “Secondly, we want to produce multiple immunotherapies from a single package, and therefore be able to stimulate the immune system in multiple different ways.”
To do this, Lu and a team including MIT postdocs Lior Nissim and Ming-Ru Wu, have built a gene circuit encoded in DNA designed to distinguish cancer cells from noncancer cells. The circuit, which can be customized to respond to different types of tumor, is based on the simple AND gates used in electronics. Such AND gates will only switch on a circuit when two inputs are present.
Cancer cells differ from normal cells in the profile of their gene expression. So the researchers developed synthetic promoters — DNA sequences designed to initiate gene expression but only in cancer cells. The circuit is delivered to cells in the affected area of the body using a virus. The synthetic promotors are then designed to bind to certain proteins that are active in tumor cells, causing the promoters to turn on. “Only when two of these cancer promoters are activated, does the circuit itself switch on,” Lu says.
This allows the circuit to target tumors more accurately than existing therapies, as it requires two cancer-specific signals to be present before it will respond. Once activated, the circuit expresses proteins designed to direct the immune system to target the tumor cells, including surface T cell engagers, which direct T cells to kill the cells. The circuit also expresses a checkpoint inhibitor designed to lift the brakes on T cell activity. When the researchers tested the circuit in vitro, they found that it was able to detect ovarian cancer cells from amongst other noncancerous ovarian cells and other cell types.
They then tested the circuit in mice implanted with ovarian cancer cells, and demonstrated that it could trigger T cells to seek out and kill the cancer cells without harming other cells around them.
Finally, the researchers showed that the circuit could be readily converted to target other cancer cells. “We identified other promoters that were selective for breast cancer, and when these were encoded into the circuit, it would target breast cancer cells over other types of cell,” Lu says.
Ultimately, they hope they will also be able to use the system to target other diseases, such as rheumatoid arthritis, inflammatory bowel disease, and other autoimmune diseases.
An MIT grad student, with help from his advisor, designed a $100 handheld muon detector that fits in shirt pocket as it detects charged particles.
At any given moment, the Earth’s atmosphere is showered with high-energy cosmic rays that have been blasted from supernovae and other astrophysical phenomena far beyond the Solar System. When cosmic rays collide with the Earth’s atmosphere, they decay into muons — charged particles that are slightly heavier than an electron.
Muons last only fractions of a second, and during their fleeting lifespan they can be found through every layer of the Earth’s atmosphere, circulating in the air around us and raining onto the surface at a rate similar to a light drizzle. A small fraction of muons can even penetrate the Earth’s surface and travel several kilometers through rock and ice.
Now physicists working in MIT's Laboratory for Nuclear Science have designed a pocket-sized cosmic ray muon detector to track these ghostly particles. The detector can be made with common electrical parts, and when turned on, it lights up and counts each time a muon passes through. The relatively simple device costs just $100 to build, making it the most affordable muon detector available today.
The researchers, led by Spencer Axani, a graduate student in MIT’s Department of Physics, have designed the detector with students in mind. They have started an outreach program called CosmicWatch, with a website that lists parts to purchase and detailed instructions on how to assemble, calibrate, and run the detector. The team estimates that an average high school student should spend about four hours building a detector for the first time, and just one hour building it a second time.
Once up and running, detectors can be carried around to measure muon rates in virtually any environment. The team has helped supply nearly 100 detectors to high school and college students, who have sent the instruments up in planes and weather balloons to measure muon rates at high altitudes. Students have also, as Axani has done, taken the detectors underground.
“You get funny looks when you take particle detectors into the subway, but we did that in Boston,” Axani says. “Since the muon rate will decrease the further down you go, we put the detectors in a subway station to measure how far underground we were.”
The researchers have published the first version of the detector design in the American Journal of Physics. Axani’s co-authors are MIT professor of physics Janet Conrad and junior Conor Kirby.
Details regarding their latest version can be found on the CosmicWatch webpage.
A University of Geneva (UNIGE) researcher has argued that models of the universe’s expansion and celestial movements can work without the concepts of dark matter and energy – a position that runs counter to a century of scientific consensus. According to a UNIGE press release, by doing away with the ideas of dark matter and dark energy – or at least the need for them to exist – the concept could contribute to “potentially solving two of astronomy’s greatest mysteries”.
The theory is the work of André Maeder, a professor in the university’s Department of Astronomy, and has recently been published in The Astrophysical Journal. Maeder goes back to basics to posit a “starting hypothesis that hasn’t been taken into account”, he says. By this, he means the scale invariance of empty space: “in other words, the empty space and its properties do not change following a dilation or a contraction”.
In the past, scientists have operated on the assumption that another Swiss astronomer, Fritz Zwicky, coined in 1933: namely, that there is more matter in the universe that we can actually see. Such an assumption, according to his and subsequent models, could account for both the universe’s ongoing expansion and the movements and speed of the stars.
The invisible substance, or matter, became known as dark matter. Its counterpart force, dark energy, is allegedly stronger than the gravitational energy discovered by Newton. However, there remained a catch: nobody could find either one. This is where Maeder’s theory comes in. After several rounds of cosmological testing of a model based on constant empty space, he found that it held true in the context of both Einsteinian relativist theories and Newton’s laws of acceleration. The model predicted the accelerated expansion of the universe without having to factor in any unseen particle or dark energy – that is, there was no missing variable in his equations.
Other tests based on the speeds reached by stars in the outer reaches of galactic clusters, and stars oscillating around the plane of the Milky Way, also pointed to explanations that relied not on dark matter but on the invariant empty space hypothesis.
For now, the ideas are more likely to inspire debate and controversy than ultimate acceptance. But for Maeder, at least, “the announcement of this model, which at last solves two of astronomy’s greatest mysteries, remains true to the spirit of science: nothing can ever be taken for granted, not in terms of experience, observation or the reasoning of human beings.”
Carbon nanotubes are extremely lightweight, electrically highly conductive, and more stable than steel. Due to their unique properties, they are ideal for numerous applications, including ultra-lightweight batteries, high-performance plastics and medical implants. However, to date, it has been difficult for science and industry to transfer the extraordinary characteristics at the nano-scale into functional industrial applications. The carbon nanotubes either cannot be combined adequately with other materials, or they lose their beneficial properties once combined.
Scientists from the Functional Nanomaterials working group at Kiel University (CAU) and the University of Trento have now developed an alternative method, with which the tiny tubes can be combined with other materials so that they retain their characteristic properties. The research results have now been published in Nature Communications.
"Although carbon nanotubes are flexible like fibre strands, they are also very sensitive to changes," explained Professor Rainer Adelung, head of the Functional Nanomaterials working group at the CAU. "With previous attempts to chemically connect them with other materials, their molecular structure also changed. This, however, made their properties deteriorate—mostly drastically."
In contrast, the approach of the research team from Kiel and Trento is based on a simple wet chemical infiltration process. The CNTs are mixed with water and dripped into an extremely porous ceramic material made of zinc oxide, which absorbs the liquid like a sponge. The dripped thread-like CNTs attach themselves to the ceramic scaffolding, and automatically form a stable layer together. The ceramic scaffolding is coated with nanotubes. This has fascinating effects, both for the scaffolding as well as for the coating of nanotubes.
On the one hand, the stability of the ceramic scaffold increases so massively that it can bear 100,000 times its own weight. "With the CNT coating, the ceramic material can hold around 7.5kg, and without it just 50g—as if we had fitted it with a close-fitting pullover made of carbon nanotubes, which provide mechanical support," said first author Fabian Schütt. "The pressure on the material is absorbed by the tensile strength of the CNT felt. Compressive forces are transformed into tensile forces."
Researchers have used a simple inkjet printer to print a "bio-ink" of cyanobacteria onto a conductive surface, creating a biophotovoltaic cell. Unlike conventional photovoltaic cells that operate only when exposed to light, the cyanobacteria can generate an electric current both in the dark and in response to light. The researchers expect that the cell may serve as an environmentally friendly power supply for low-power devices such as biosensors, and can even be scaled up to print a bioenergy wallpaper.
The scientists, at Imperial College London and University of Cambridge, have published a paper on the new biophotovoltaic cell in a recent issue of Nature Communications. "Our biophotovoltaic device is biodegradable and in the future could serve as a disposable solar panel and battery that can decompose in our composts or gardens," coauthor Marin Sawa at University of Arts London and Imperial College London told Phys.org. "Cheap, accessible, environmentally friendly, biodegradable batteries without any heavy metals and plastics—this is what we and our environment really need but don't have just yet, and our work has shown that it is possible to have that."
In general, biophotovoltaic cells contain some type of cyanobacteria or algae that is phototrophic, meaning it converts light into energy. However, even in the dark these organisms continue to generate some energy by metabolizing their internal storage reserves. So when the organisms are connected to a non-biological electrode, they can function as either a "bio solar panel" when exposed to light or a "solar bio-battery" in the dark.
Currently one of the biggest challenges facing biophotovoltaic cells is producing them on a large scale. Typically, the organisms are deposited onto an electrode surface from a bulky liquid reservoir. In the new study, the researchers demonstrated that inkjet printing can be used to print both the carbon nanotube electrode surface and the cyanobacteria on top of it, while allowing the bacteria to remain fully viable. This approach not only allows the cells to be fabricated quickly, but the set-up is also more compact and allows for greater precision in cell design.
With these advantages, the inkjet-printed biophotovoltaic cells can generate a maximum current density that is 3-4 times higher than cells fabricated using conventional methods. To demonstrate, the researchers showed that nine connected cells can power a digital clock or generate flashes of light from an LED, illustrating the ability to produce short bursts of relatively high power. The researchers also showed that the cells can generate a continuous power output over the course of a 100-hour period consisting of light and dark cycles.
In the future, the researchers plan to develop thin-film biophotovoltaic (BPV) panels and also explore potential applications as integrated power supplies in the areas of point-of-care medical diagnostics and environmental monitoring, both of which benefit from disposable, environmentally friendly biosensors. Another potential application is a bioenergy wallpaper.
"The bioenergy wallpaper is a scaled-up application of our BPV system," Sawa said. "The wallpaper will have carbon-based conductive patterns with electron-producing cyanobacteria. It turns an interior surface into an energy harvester to drive low-power applications like LED lights and/or biosensors, which can, for example, monitor indoor air quality."
Researchers recently discovered what was thought to be a distinct species of butterfly is actually the female of a species known to science for more than a century.
An international team of nine butterfly researchers from the U.S., Brazil, the U.K., Peru and Germany used DNA sequence data to associate the female sunburst cerulean-satyr, or Caeruleuptychia helios, an Amazonian brush-footed butterfly, with its male counterpart.
Males and females of this group look dramatically different from each other, a phenomenon known as sexual dimorphism, and the species was named and described in 1911 based on the brilliantly iridescent blue males. Rarer than the male, the brown female was considered another species and was recently named and placed in a different genus, Magneuptychia keltoumae.
A study correcting the classification error was published today in Insect Systematics and Evolution.
"Our study will serve as the basis for developing a firm understanding of true species diversity of this group and of Neotropical butterflies in general," said Shinichi Nakahara, the study's lead author and a research associate at the Florida Museum of Natural History's McGuire Center for Lepidoptera and Biodiversity at the University of Florida. "These findings are extremely valuable at a time when the biodiversity of the Neotropics is threatened since it will be impossible to recognize and document the region's unique elements of biodiversity after they are gone."
As part of the project, DNA bar codes—short, diagnostic gene sequences—were collected and analyzed for more than 300 species across the euptychiines group of butterflies.
Life on Earth might have originated from tiny organisms brought to our planet in streams of fast-moving space dust, according to a new study.
Fast-moving flows of interplanetary dust that continually bombard our planet's atmosphere could deliver tiny organisms from far-off worlds, or send Earth-based organisms to other planets, according to the research. The dust streams could collide with biological particles in Earth's atmosphere with enough energy to knock them into space, a scientist has suggested. Such an event could enable bacteria and other forms of life to make their way from one planet in the solar system to another and perhaps beyond.
The finding suggests that large asteroid impacts may not be the sole mechanism by which life could transfer between planets, as was previously thought.
The research from the University of Edinburgh calculated how powerful flows of space dust -- which can move at up to 70 km a second -- could collide with particles in our atmospheric system.
It found that small particles existing at 150 km or higher above Earth's surface could be knocked beyond the limit of Earth's gravity by space dust and eventually reach other planets. The same mechanism could enable the exchange of atmospheric particles between distant planets.
Some bacteria, plants and small animals called tardigrades are known to be able to survive in space, so it is possible that such organisms -- if present in Earth's upper atmosphere -- might collide with fast-moving space dust and withstand a journey to another planet.
The study, published in Astrobiology, was partly funded by the Science and Technology Facilities Council. Professor Arjun Berera, from the University of Edinburgh's School of Physics and Astronomy, who led the study, said: "The proposition that space dust collisions could propel organisms over enormous distances between planets raises some exciting prospects of how life and the atmospheres of planets originated. The streaming of fast space dust is found throughout planetary systems and could be a common factor in proliferating life."
Researchers have shown how to write any magnetic pattern desired onto nanowires, which could help computers mimic how the brain processes information. Much current computer hardware, such as hard drives, use magnetic memory devices. These rely on magnetic states -- the direction microscopic magnets are pointing -- to encode and read information.
Exotic magnetic states -- such as a point where three south poles meet -- represent complex systems. These may act in a similar way to many complex systems found in nature, such as the way our brains process information. Computing systems that are designed to process information in similar ways to our brains are known as 'neural networks'. There are already powerful software-based neural networks -- for example one recently beat the human champion at the game 'Go' -- but their efficiency is limited as they run on conventional computer hardware.
Now, researchers from Imperial College London have devised a method for writing magnetic information in any pattern desired, using a very small magnetic probe called a magnetic force microscope. With this new writing method, arrays of magnetic nanowires may be able to function as hardware neural networks -- potentially more powerful and efficient than software-based approaches. The team, from the Departments of Physics and Materials at Imperial, demonstrated their system by writing patterns that have never been seen before. They published their results today in Nature Nanotechnology.
Dr Jack Gartside, first author from the Department of Physics, said: "With this new writing method, we open up research into 'training' these magnetic nanowires to solve useful problems. If successful, this will bring hardware neural networks a step closer to reality." As well as applications in computing, the method could be used to study fundamental aspects of complex systems, by creating magnetic states that are far from optimal (such as three south poles together) and seeing how the system responds.
To get content containing either thought or leadership enter:
To get content containing both thought and leadership enter:
To get content containing the expression thought leadership enter:
You can enter several keywords and you can refine them whenever you want. Our suggestion engine uses more signals but entering a few keywords here will rapidly give you great content to curate.
Your new post is loading...
