Facebook’s parent company Meta has unveiled a text-to-video AI model called Make-A-Video. Like DALL-E, the system works based on text prompts, generating short five-second videos based on users’ descriptions.
Make-A-Video is a new AI system that lets people turn text prompts into brief, high-quality video clips. Make-A-Video builds on Meta AI’s recent progress in generative technology research and has the potential to open new opportunities for creators and artists. The system learns what the world looks like from paired text-image data and how the world moves from video footage with no associated text. As part of our continued commitment to open science, Meta AI is sharing details in a recent research paper.
Generative AI research is pushing creative expression forward by giving people tools to quickly and easily create new content. With just a few words or lines of text, Make-A-Video can bring imagination to life and create one-of-a-kind videos full of vivid colors, characters, and landscapes. The system can also create videos from images or take existing videos and create new ones that are similar.
Make-A-Video follows Meta AI's announcement earlier this year of Make-A-Scene, a multimodal generative AI method that gives people more control over the AI generated content they create. With Make-A-Scene, we demonstrated how people can create photorealistic illustrations and storybook-quality art using words, lines of text, and freeform sketches.
The Meta AI designers want to be thoughtful about how they build new generative AI systems like this. Make-A-Video uses publicly available datasets, which adds an extra level of transparency to the research. They are openly sharing this generative AI research and results with the community for their feedback, and will continue to use a responsible AI framework to refine and evolve their approach to this emerging technology.
Learn more about Make-A-Video by visiting Meta AI's site and read the paper.
Millions of people are prescribed statins such as Lipitor, Crestor or generic formulations to lower their cholesterol. But lots of people are hesitant to start the medication. Some people fret over potential side effects such as leg cramps, which may be — or may not be — linked to the drug. As an alternative, dietary supplements, often marketed to promote heart health, including fish oil and other omega-3 supplements (omega-3s are essential fatty acids found in fish and flaxseed), are growing in popularity.
So, which is most effective? Researchers at the Cleveland Clinic set out to answer this question by comparing statins to supplements in a clinical trial. They tracked the outcomes of 190 adults, ages 40 to 75. Some participants were given a 5 mg daily dose of rosuvastatin, a statin that is sold under the brand name Crestor for 28 days. Others were given supplements, including fish oil, cinnamon, garlic, turmeric, plant sterols or red yeast rice for the same period.
"What we found was that rosuvastatin lowered LDL cholesterol by almost 38% and that was vastly superior to placebo and any of the six supplements studied in the trial," study author Luke Laffin, M.D. of the Cleveland Clinic's Heart, Vascular & Thoracic Institute told NPR. He says this level of reduction is enough to lower the risk of heart attacks and strokes. The findings are published in theJournal of the American College of Cardiology.
"Clearly, statins do what they're intended to do," the study's senior author Steve Nissen, a cardiologist and chief academic officer of the Heart, Vascular & Thoracic Institute at Cleveland Clinic told NPR. By comparison, he says this research shows that supplements are not effective. "They do not promote heart health. They do not improve levels of the bad cholesterol." Nissen says supplements can be expensive compared to statin medications. Depending on insurance, he says people may pay less than $5 a month out-of-pocket for rosuvastatin.
"Statins are the most effective heart attack and stroke prevention drugs that we have really ever seen," says Michael Honigberg, a cardiologist and researcher at Massachusetts General Hospital who is not affiliated with the new study. He says the new findings add to an already large body of evidence showing statins lower LDL cholesterol, and he's not surprised to see that the supplements were not as effective.
However, he says, not everyone with a family history of heart disease or slightly elevated cholesterol should be on a statin drug. The American College of Cardiology and American Heart Association developed some prescription guidelines. Typically, if a person's LDL cholesterol (bad cholesterol) is 190 or higher, they're often advised to start a statin. Health care professionals use a risk calculator to estimate a person's risk of having a heart attack or stroke over the next 10 years. If the risk is high enough, based on factors including age, blood pressure and smoking status, then a statin may be recommended.
Recent reports suggest that cell-surface and intracellular immune receptors function synergistically to activate a robust defense against pathogens, but whether they co-evolve is unclear. Plant geneticists now determined the numbers of cell-surface and intracellular immune receptors in 350 species. Surprisingly, the number of receptor genes that are predicted to encode cell-surface and intracellular immune receptors is strongly correlated. The scientists suggest this is consistent with mutual potentiation of immunity initiated by cell-surface and intracellular receptors being reflected in the concerted co-evolution of the size of their repertoires across plant species. Comparative genomic analysis of 350 plant species reveals that cell-surface and intracellular immune receptor gene families co-expand or co-contract. This suggests an evolutionary relationship between the two branches of the plant immune system.
Plants have evolved a two-tier immune system that recognizes and activates defense against a large variety of pathogens1,2. Cell-surface pattern-recognition receptors (PRRs) recognize apoplastic and usually conserved pathogen-associated molecular patterns (PAMPs) and activate pattern-triggered immunity (PTI). Virulent pathogens secrete effector molecules into plant cells that suppress PTI and promote infection. Intracellular nucleotide-binding leucine-rich repeat (NLR) receptors recognize effectors and activate effector-triggered immunity (ETI). Although PTI and ETI were envisaged as two independent immune systems1, emerging evidence suggests they are inter-dependent and share multiple signaling components3,4,5,6. Thus, PTI and ETI function synergistically to provide robust immunity against pathogens. As PRRs and NLRs are functionally inter-dependent, the authors of this Brief Communication investigated whether the sizes of these two receptor gene families are somewhat correlated.
Plant PRR proteins are structurally diverse but are usually receptor-like kinases (RLKs) or receptor-like proteins (RLPs). RLKs carry extracellular ectodomains and cytosolic kinase domains, while RLPs lack cytosolic kinase domains altogether. RLKs carry multiple types of extracellular domains, such as leucine-rich repeats (LRRs), lectins and lysM motifs (LysMs)7. LRR-domain-containing RLKs (LRR-RLKs) and RLPs (LRR-RLPs) are the largest RLK- and RLP-gene families in plants8,9. LRR-RLKs can be further classified into 20 subgroups, with each subgroup involved in different biological processes10 (Extended Data Fig. 1). For example, the BAK1 (BRI1-ASSOCIATED RECEPTOR KINASE) family of proteins functions as PRR co-receptors and belongs to LRR-RLK-II (ref. 11). Members of the LRR-RLK-XI are involved in recognition of self-peptides12,13. Members of LRR-RLK-XII on the other hand, such as FLAGELLIN-SENSITIVE 2 (FLS2), EF-TU RECEPTOR (EFR) and Xa21 (refs. 2,7), are involved in detecting pathogen-derived molecules (Extended Data Fig. 1). NLRs are intracellular receptors that carry NB-ARC domains with C-terminal LRR domains and N-terminal domains, usually comprising either coiled-coil (CC), Toll/interleukin-1 receptor/resistance protein (TIR) or RPW8-like coiled-coil (RPW8) domains (hence, CC-NLRs (or CNLs), TIR-NLRs (TNLs) and RPW8-NLRs (RNLs))14,15.
To investigate expansion or contraction of genes that encode PRR and NLR proteins, the scientists identified these gene families in annotated proteomes from 350 publicly available genomes. These genomes included 26 algal species, 5 bryophyte species, 10 gymnosperms and 300 angiosperms (13 basal angiosperms, 79 monocots and 208 eudicots). Assembled genome sizes of these organisms range from 13 Mb to 27.6 Gb, with annotated protein counts ranging from ~5,000 to ~300,000. To ensure consistency, the authors of the study used the same pipeline to obtain primary transcripts and identify LRR-RLKs, LRR-RLPs, LysM-RLK, LysM-RLP and NB-ARCs from each of these genomes.
Researchers at DeepMind in London have shown that artificial intelligence (AI) can find shortcuts in a fundamental type of mathematical calculation, by turning the problem into a game and then leveraging the machine-learning techniques that another of the company’s AIs used to beat human players in games such as Go and chess.
The AI discovered algorithms that break decades-old records for computational efficiency, and the team’s findings, published on 5 October 2022 in Nature1, could open up new paths to faster computing in some fields. “It is very impressive,” says Martina Seidl, a computer scientist at Johannes Kepler University in Linz, Austria. “This work demonstrates the potential of using machine learning for solving hard mathematical problems.”
Algorithms chasing algorithms
Advances in machine learning have allowed researchers to develop AIs that generate language, predict the shapes of proteins2 or detect hackers. Increasingly, scientists are turning the technology back on itself, using machine learning to improve its own underlying algorithms.
The AI that DeepMind developed — called AlphaTensor — was designed to perform a type of calculation called matrix multiplication. This involves multiplying numbers arranged in grids — or matrices — that might represent sets of pixels in images, air conditions in a weather model or the internal workings of an artificial neural network. To multiply two matrices together, the mathematician must multiply individual numbers and add them in specific ways to produce a new matrix. In 1969, mathematician Volker Strassen found a way to multiply a pair of 2 × 2 matrices using only seven multiplications3, rather than eight, prompting other researchers to search for more such tricks.
DeepMind’s approach uses a form of machine learning called reinforcement learning, in which an AI ‘agent’ (often a neural network) learns to interact with its environment to achieve a multistep goal, such as winning a board game. If it does well, the agent is reinforced — its internal parameters are updated to make future success more likely.
AlphaTensor also incorporates a game-playing method called tree search, in which the AI explores the outcomes of branching possibilities while planning its next action. In choosing which paths to prioritize during tree search, it asks a neural network to predict the most promising actions at each step. While the agent is still learning, it uses the outcomes of its games as feedback to hone the neural network, which further improves the tree search, providing more successes to learn from.
Each game is a one-player puzzle that starts with a 3D tensor — a grid of numbers — filled in correctly. AlphaTensor aims to get all the numbers to zero in the fewest steps, selecting from a collection of allowable moves. Each move represents a calculation that, when inverted, combines entries from the first two matrices to create an entry in the output matrix. The game is difficult, because at each step the agent might need to select from trillions of moves. “Formulating the space of algorithmic discovery is very intricate,” co-author Hussein Fawzi, a computer scientist at DeepMind, said at a press briefing, but “even harder is, how can we navigate in this space”.
AI can be utilized to help students find more success in mathematics. As shown in this article, the limits to AI are endless. Leveraging AI in the mathematics classroom is a great way to improve outcomes.
A major question in Mars history is whether the planet had an ocean in its northern hemisphere.
New research provides evidence for a northern ocean or large sea based on stratigraphic analysis of sedimentary basin fill exposed at Aeolis Dorsa, a trough-like region of Mars about 500 km wide and 900 km long, surrounded east and west by 2 km-high plains and to the south by steep slopes up to the southern hemisphere.
“What immediately comes to mind as one the most significant points here is that the existence of an ocean of this size means a higher potential for life. It also tells us about the ancient climate and its evolution,” said Penn State geoscientist Benjamin Cardenas. “Based on these findings, we know there had to have been a period when it was warm enough and the atmosphere was thick enough to support this much liquid water at one time. There has long been debate in the scientific community about whether Mars had an ocean in its low-elevation northern hemisphere.”
Using topography data, Dr. Cardenas and his colleague, Caltech’s Dr. Michael Lamb, were able to show definitive evidence of a roughly 3.5-billion-year-old shoreline with substantial sedimentary accumulation, at least 900 m thick, that covered hundreds of thousands of square kilometers. “The big, novel thing that we did in this paper was think about Mars in terms of its stratigraphy and its sedimentary record,” Dr. Cardenas said. “On Earth, we chart the history of waterways by looking at sediment that is deposited over time. We call that stratigraphy, the idea that water transports sediment and you can measure the changes on Earth by understanding the way that sediment piles up. That’s what we’ve done here — but it’s Mars.”
The authors discovered over 6,500 km of fluvial ridges and grouped them into 20 systems to show that the ridges are likely eroded river deltas or submarine-channel belts, the remnants of an ancient Martian shoreline. Elements of rock formations, such as ridge-system thicknesses, elevations, locations and possible sedimentary flow directions helped the team understand the evolution of the region’s paleogeography. The area that was once ocean is now known as Aeolis Dorsa and contains the densest collection of fluvial ridges on the planet.
“The rocks in Aeolis Dorsa capture some fascinating information about what the ocean was like,” Dr. Cardenas said. “It was dynamic. The sea level rose significantly. Rocks were being deposited along its basins at a fast rate. There was a lot of change happening here.” The team’s work appeared this month in the Journal of Geophysical Research: Planets.
Scientists with the University of Chicago have discovered a way to create a material that can be made like a plastic, but conducts electricity more like a metal.
The research, published Oct. 26th 2022 in Nature, shows how to make a kind of material in which the molecular fragments are jumbled and disordered, but can still conduct electricity extremely well. This goes against all of the rules we know about for conductivity -- to a scientist, it's kind of seeing a car driving on water and still going 70 mph. But the finding could also be extraordinarily useful; if you want to invent something revolutionary, the process often first starts with discovering a completely new material.
"In principle, this opens up the design of a whole new class of materials that conduct electricity, are easy to shape, and are very robust in everyday conditions," said John Anderson, an associate professor of chemistry at the University of Chicago and the senior author on the study. "Essentially, it suggests new possibilities for an extremely important technological group of materials," said Jiaze Xie (PhD'22, now at Princeton), the first author on the paper.
'There isn't a solid theory to explain this'
Conductive materials are absolutely essential if you're making any kind of electronic device, whether it be an iPhone, a solar panel, or a television. By far the oldest and largest group of conductors is the metals: copper, gold, aluminum. Then, about 50 years ago, scientists were able to create conductors made out of organic materials, using a chemical treatment known as "doping," which sprinkles in different atoms or electrons through the material. This is advantageous because these materials are more flexible and easier to process than traditional metals, but the trouble is they aren't very stable; they can lose their conductivity if exposed to moisture or if the temperature gets too high.
But fundamentally, both of these organic and traditional metallic conductors share a common characteristic. They are made up of straight, closely packed rows of atoms or molecules. This means that electrons can easily flow through the material, much like cars on a highway. In fact, scientists thought a material had to have these straight, orderly rows in order to conduct electricity efficiently. Then Xie began experimenting with some materials discovered years ago, but largely ignored. He strung nickel atoms like pearls into a string of of molecular beads made of carbon and sulfur, and began testing.
To the scientists' astonishment, the material easily and strongly conducted electricity. What's more, it was very stable. "We heated it, chilled it, exposed it to air and humidity, and even dripped acid and base on it, and nothing happened," said Xie. That is enormously helpful for a device that has to function in the real world. But to the scientists, the most striking thing was that the molecular structure of the material was disordered. "From a fundamental picture, that should not be able to be a metal," said Anderson. "There isn't any solid theory to explain this."
Recent breakthroughs in text-to-image synthesis have been driven by diffusion models trained on billions of image-text pairs. Adapting this approach to 3D synthesis would require large-scale datasets of labeled 3D assets and efficient architectures for denoising 3D data, neither of which currently exist. These limitations can be avoided by using a pretrained 2D text-to-image diffusion model to perform text-to-3D synthesis.
Dreamfusion introduces a loss based on probability density distillation that enables the use of a 2D diffusion model as a prior for optimization of a parametric image generator. Using this loss in a DeepDream-like procedure, allows to optimize a randomly-initialized 3D model (a Neural Radiance Field, or NeRF) via gradient descent such that its 2D renderings from random angles achieve a low loss. The resulting 3D model of the given text can be viewed from any angle, relit by arbitrary illumination, or composited into any 3D environment. This approach requires no 3D training data and no modifications to the image diffusion model, demonstrating the effectiveness of pretrained image diffusion models as priors.
This phylogenetic tree represents the clade distribution of global SARS-CoV-2 viral genomes since the beginning of the pandemic. In the center is the reference Wuhan-1 strain, with outside samples now separated by nucleotide divergence. The last few months have led to an explosion of new variants with further mutations in the viral spike resulting in further antibody evasion, and also carrying changes in many other viral proteins. Some recently sequenced viral genomes from samples collected in Southeast Asia (not shown in this figure) display over a hundred nucleotides changes when compared to Wuhan-1. The graph was generated with nextstrain.org
The annual Antarctic ozone hole reached an average area of 8.9 million square miles (23.2 million square kilometers) between Sept. 7 and Oct. 13, 2022. This depleted area of the ozone layer over the South Pole was slightly smaller than last year and generally continued the overall shrinking trend of recent years.
“Over time, steady progress is being made, and the hole is getting smaller,” said Paul Newman, chief scientist for Earth sciences at NASA’s Goddard Space Flight Center in Greenbelt, Maryland. “We see some wavering as weather changes and other factors make the numbers wiggle slightly from day to day and week to week. But overall, we see it decreasing through the past two decades. The elimination of ozone-depleting substances through the Montreal Protocol is shrinking the hole.”
The ozone layer – the portion of the stratosphere that protects our planet from the Sun’s ultraviolet rays – thins to form an “ozone hole” above the South Pole every September. Chemically active forms of chlorine and bromine in the atmosphere, derived from human-produced compounds, attach to high-altitude polar clouds each southern winter. The reactive chlorine and bromine then initiate ozone-destroying reactions as the Sun rises at the end of Antarctica’s winter.
Researchers at NASA and NOAA detect and measure the growth and breakup of the ozone hole with instruments aboard the Aura, Suomi NPP, and NOAA-20 satellites. On Oct. 5, 2022, those satellites observed a single-day maximum ozone hole of 10.2 million square miles (26.4 million square kilometers), slightly larger than last year.
The virus that causes COVID-19 is a prolific sack of genes that targets not just humans but nonhuman animals as well. And just as humans and animals can infect one another, animals can also infect other animals, says Amélie Desvars-Larrive, an epidemiologist at the University of Veterinary Medicine in Vienna. Scientists have learned a lot about how COVID spreads in humans but less about how it spreads between animals. To make it easier to study the connections among humans, animals and the virus, Desvars-Larrive and a team of researchers gathered scattered reports of COVID-infected mammals from all over the world to create a public database. Understanding how the virus spreads between nonhuman mammals, and then between those mammals and humans, can help us better manage the current pandemic—and prepare for the next one.
“We can’t continue to focus on humans, to have an anthropocentric point of view on this pandemic,” Desvars-Larrive says. COVID has proved highly contagious among many mammal species: Infections have run rampant among captive mink, and fur farmers have had to kill their entire stocks to stop it. Deer appear particularly susceptible to the virus. Many cat species, both big and small, seem to get it as well. Barbara Han, a disease ecologist at the Cary Institute of Ecosystem Studies in Millbrook, N.Y., who was not involved in the database project, says having this kind of information in one place—rather than split among multiple sources managed by different organizations and government agencies—will likely make it faster and easier for her team to predict how the virus behaves. The database is growing as more animals are tested and reports shared, and scientists hope it will help them to track animal-to-animal COVID infections as well as transmission between animals and humans. Han says that beyond just tallying individual infected species, the database will make it easier for scientists to study how the COVID-causing SARS-CoV-2 virus affects mammal communities and entire ecosystems. “People are fascinated that this pathogen is now in [so many] animals and what it might mean for us,” Han says. “If we don’t have good information about which animals have it now, we can’t get those answers.”
Built to help scientists understand how dust affects climate, the Earth Surface Mineral Dust Source Investigation can also pinpoint emissions of the potent greenhouse gas.
NASA’s Earth Surface Mineral Dust Source Investigation (EMIT) mission is mapping the prevalence of key minerals in the planet’s dust-producing deserts – information that will advance our understanding of airborne dust’s effects on climate. But EMIT has demonstrated another crucial capability: detecting the presence of methane, a potent greenhouse gas.
In the data EMIT has collected since being installed on the International Space Station in July, the science team has identified more than 50 “super-emitters” in Central Asia, the Middle East, and the Southwestern United States. Super-emitters are facilities, equipment, and other infrastructure, typically in the fossil-fuel, waste, or agriculture sectors, that emit methane at high rates.
“Reining in methane emissions is key to limiting global warming.
This exciting new development will not only help researchers better pinpoint where methane leaks are coming from, but also provide insight on how they can be addressed – quickly,” said NASA Administrator Bill Nelson. “The International Space Station and NASA’s more than two dozen satellites and instruments in space have long been invaluable in determining changes to the Earth’s climate. EMIT is proving to be a critical tool in our toolbox to measure this potent greenhouse gas – and stop it at the source.”
Methane absorbs infrared light in a unique pattern – called a spectral fingerprint – that EMIT’s imaging spectrometer can discern with high accuracy and precision. The instrument can also measure carbon dioxide. The new observations stem from the broad coverage of the planet afforded by the space station’s orbit, as well as from EMIT’s ability to scan swaths of Earth’s surface dozens of miles wide while resolving areas as small as a soccer field.
“These results are exceptional, and they demonstrate the value of pairing global-scale perspective with the resolution required to identify methane point sources, down to the facility scale,” said David Thompson, EMIT’s instrument scientist and a senior research scientist at NASA’s Jet Propulsion Laboratory in Southern California, which manages the mission. “It’s a unique capability that will raise the bar on efforts to attribute methane sources and mitigate emissions from human activities.”
Relative to carbon dioxide, methane makes up a fraction of human-caused greenhouse-gas emissions, but it’s estimated to be 80 times more effective, ton for ton, at trapping heat in the atmosphere in the 20 years after release. Moreover, where carbon dioxide lingers for centuries, methane persists for about a decade, meaning that if emissions are reduced, the atmosphere will respond in a similar timeframe, leading to slower near-term warming.
Identifying methane point sources can be a key step in the process. With knowledge of the locations of big emitters, operators of facilities, equipment, and infrastructure giving off the gas can quickly act to limit emissions.
TESS is conducting a survey of about 200,000 of the brightest stars near the sun with the aim of searching for transiting exoplanets. So far, it has identified nearly 6,000 candidate exoplanets (TESS Objects of Interest, or TOI), of which 266 have been confirmed so far.
Now, a group of astronomers led by Silvano Desidera of the Astronomical Observatory of Padova, has recently confirmed another TOI monitored by TESS. They report that a transit signal has been identified in the light curve of a bright K-dwarf star—TOI-179 (other designations HD 18599 and HIP 13754). The planetary nature of this signal was confirmed by follow-up observations using the High Accuracy Radial velocity Planet Searcher (HARPS) and Spectro-Polarimetric High-contrast Exoplanet REsearch (SPHERE) instruments.
"As part of on-going efforts to validate and characterize young transiting exoplanets identified by TESS, we present in this paper our analysis for the system observed around the star HD 18599 = HIP 13754, a bright (V=8.99 mag) and active K dwarf, also known as TESS Object of Interest (TOI)-179," the researchers wrote in the paper.
The newfound alien world, designated TOI-179 b, is about 2.62 times larger than the Earth and 24 times more massive than our planet, which yields a relatively high mean density of some 7.4 g/cm3. The exoplanet orbits its host every 4.14 days, at a distance of 0.048 AU from it, on a significantly eccentric orbit.
Furthermore, SPHERE observations identified another object in the TOI-179 system with an estimated mass of about 83 Jupiter masses—therefore, at the boundary between brown dwarfs and very low-mass stars. The object received designation HD 18599 B has a relatively small projected separation from the parent star—some 3.3 AU.
The host star TOI-179 is of spectral type K2V, has a radius of approximately 0.76 solar radii, while its mass was measured to be 0.83 solar masses. The star is estimated to be around 400 million years old and its effective temperature is at a level of 5,145 K.
Summing up the results, the authors of the study underlined the uniqueness of the TOI-179 system taking into account the properties of its components. "The TOI-179 system represents a high-merit laboratory for our understanding of the physical evolution of planets and other low-mass objects and of how the planet properties are influenced by dynamical effects and interactions with the parent star," the researchers concluded.
Humanity might not "speak extraterrestrial." The SETI Post-Detection Hub wants to coordinate how we respond to alien life.
What happens when humanity's biggest sci-fi dream comes true and intelligent extraterrestrial life makes contact with Earth? A group of scientists is formally working on answering that complicated question. Recently, the University of St. Andrews in the UK announced a new research hub that will bring together an international team of experts to help prepare humanity for alien contact. The SETI Post-Detection Hub will develop a framework for "impact assessments, protocols, procedures and treaties designed to allow humanity to respond responsibly" should we detect intelligent life from beyond our planet. SETI is short for "Search for Extra-Terrestrial Intelligence."
St. Andrews SETI specialist John Elliott is the coordinator for the project, which will draw on experts from both the sciences and the humanities. The group will investigate topics ranging from message deciphering to space law to how to communicate discoveries to the public. The discovery of intelligent life may not be as obvious as aliens popping over to our planet for a visit. More likely, it might come in the form of distant communications signals that would need to be analyzed and interpreted. So far, SETI efforts have come up empty. Case in point: An Australian telescope recently scanned over 10 million stars for signs of alien tech and found nothing.
The SETI Post-Detection Hub hopes to fill a void in the discussion around humanity's potential response to the discovery of alien life. The SETI Institute, a nonprofit research organization devoted to the search for alien life, had issued a "Declaration of Principles Concerning the Conduct of the Search for Extraterrestrial Intelligence."
The hub's work is just beginning, so it hasn't offered any recommendations yet, but there is a sense of urgency to the project. "Will we ever get a message from E.T.? We don't know," said Elliot. "We also don't know when this is going to happen. But we do know that we cannot afford to be ill prepared -- scientifically, socially, and politically rudderless -- for an event that could turn into reality as early as tomorrow and which we cannot afford to mismanage."
Traditionally, 3D indoor scene reconstruction from posed images happens in two phases: per image depth estimation, followed by depth merging and surface reconstruction. Recently, a family of methods have emerged that perform reconstruction directly in final 3D volumetric feature space. While these methods have shown impressive reconstruction results, they rely on expensive 3D convolutional layers, limiting their application in resource-constrained environments. In this work, researchers instead go back to the traditional route, and show how focusing on high quality multi-view depth prediction leads to highly accurate 3D reconstructions using simple off-the-shelf depth fusion. They propose a simple state-of-the-art multi-view depth estimator with two main contributions: 1) a carefully-designed 2D CNN which utilizes strong image priors alongside a plane-sweep feature volume and geometric losses, combined with 2) the integration of keyframe and geometric metadata into the cost volume which allows informed depth plane scoring. This method achieves a significant lead over the current state-of-the-art for depth estimation and close or better for 3D reconstruction on ScanNet and 7-Scenes, yet still allows for online real-time low-memory reconstruction. SimpleRecon is fast. The batch size one performance is 70ms per frame. This makes accurate reconstruction via fast depth fusion possible!
SimpleRecon: 3D Reconstruction Without 3D Convolutions Mohamed Sayed, John Gibson, Jamie Whatson, Victor Adrian Prisacariu, Michael Firman, and Clément Godard ECCV 2022
Like modern HIV, ancient human endogenous retroviruses (HERVs) had to insert their genetic material into their host’s genome to replicate.
Viruses insert their genomes into their hosts in the form of a provirus. There are around 30 different kinds of HERVs in people today, amounting to over 60,000 proviruses in the human genome.
They demonstrate the long history of the many pandemics humanity has been subjected to over the course of evolution.
Scientists think these viruses once widely infected the population, since they have become fixed in not only the human genome but also in chimpanzee, gorilla and other primate genomes. Research has demonstrated that HERV genes are active in diseased tissue, such as tumors, as well as during human embryonic development. But how active HERV genes are in healthy tissue was still largely unknown.
To answer this question, Aidan Burn, a Ph.D. candidate at Tufts University, and his colleagues decided to focus on one group of HERVs known as HML-2. This group is the most recently active of the HERVs, having gone extinct less than 5 million years ago. Even now, some of its proviruses within the human genome still retain the ability to make viral proteins.
The researchers examined the genetic material in a database containing over 14,000 donated tissue samples from all across the body. They looked for sequences that matched each HML-2 provirus in the genome and found 37 different HML-2 proviruses that were still active. All 54 tissue samples they analyzed had some evidence of activity of one or more of these proviruses. Furthermore, each tissue sample also contained genetic material from at least one provirus that could still produce viral proteins.
Role of HERVs in Human Health and Disease
The fact that thousands of pieces of ancient viruses still exist in the human genome and can even create protein has drawn a considerable amount of attention from researchers, particularly since related viruses still active today can cause breast cancer and AIDS-like disease in animals.
Whether the genetic remnants of human endogenous retroviruses can cause disease in people is still under study. Researchers have spotted virus-like particles from HML-2 in cancer cells, and the presence of HERV genetic material in diseased tissue has been associated with conditions such as Lou Gehrig’s disease, or amyotrophic lateral sclerosis, as well as multiple sclerosis and even schizophrenia.
The new study adds a new angle to these data by showing that HERV genes are present even in healthy tissue. This means that the presence of HERV RNA may not be enough to connect the virus to a disease. Importantly, it also means that HERV genes or proteins may no longer be good targets for drugs.
HERVs have been explored as a target for a number of potential drugs, including antiretroviral medication, antibodies for breast cancer and T-cell therapies for melanoma. Treatments using HERV genes as a cancer biomarker will also need to take into account their activity in healthy tissue. On the other hand, the study also suggests that HERVs could even be beneficial to people.
The most famous HERV embedded in human and animal genomes, syncytin, is a gene derived from an ancient retrovirus that plays an important role in the formation of the placenta.
Pregnancy in all mammals is dependent on the virus-derived protein coded in this gene. Similarly, mice, cats and sheep also found a way to use endogenous retroviruses to protect themselves against the original ancient virus that created them.
While these embedded viral genes are unable to use their host’s machinery to create a full virus, enough of their damaged pieces circulate in the body to interfere with the replication cycle of their ancestral virus if the host encounters it. Scientists theorize that one HERV may have played this protective role in people millions of years ago.
The study highlights a few more HERVs that could have been claimed or co-opted by the human body much more recently for this same purpose.
Microbial molecules from soil, seawater and human bodies are among the planet’s least understood proteins. The ESM Metagenomic Atlas database contains structure predictions for 617 million proteins.
When London-based Deep Mind unveiled predicted structures for some 220 million proteins this year, it covered nearly every protein from known organisms in DNA databases. Now, another tech giant is filling in the dark matter of our protein universe.
Researchers at Meta (formerly Facebook) have used artificial intelligence (AI) to predict the structures of some 600 million proteins from bacteria, viruses and other microbes that haven’t been characterized. ‘It will change everything’: DeepMind’s AI makes gigantic leap in solving protein structures
“These are the structures we know the least about. These are incredibly mysterious proteins. I think they offer the potential for great insight into biology,” says Alexander Rives, the research lead for Meta AI’s protein team.
The team generated the predictions — described in a November 2022 preprint — using a ‘large language model’, a type of AI that are the basis for tools that can predict text from just a few letters or words.
Normally language models are trained on large volumes of text. To apply them to proteins, Rives and his colleagues fed them sequences to known proteins, which can be expressed by a chains of 20 different amino acids, each represented by a letter. The network then learned to ‘autocomplete’ proteins with a proportion of amino acids obscured.
Protein ‘autocomplete’ This training imbued the network with an intuitive understanding of protein sequences, which hold information about their shapes, says Rives. A second step — inspired by DeepMind’s pioneering protein structure AI AlphaFold — combines such insights with information about the relationships between known protein structures and sequences, to generate predicted structures from protein sequences.
Meta’s network, called ESMFold, isn’t quite as accurate as AlphaFold, Rives’ team reported earlier this summer2, but it is about 60 times faster at predicting structures, he says. “What this means is that we can scale structure prediction to much larger databases.”
As a test case, they decided to wield their model on a database of bulk-sequenced ‘metagenomic’ DNA from environmental sources including soil, seawater, the human gut, skin and other microbial habitats. The vast majority of the DNA entries — which encode potential proteins — come from organisms that have never been cultured and are unknown to science.
In total, the Meta team predicted the structures of more than 617 million proteins. The effort took just 2 weeks (AlphaFold can take minutes to generate a single prediction). The predictions are freely available for anyone to use, as is the code underlying the model, says Rives. Of these 617 million predictions, the model deemed more than one-third to be high quality, such that researchers can have confidence that the overall protein shape is correct and, in some cases, can discern finer atomic-level details. Millions of these structures are entirely novel, and unlike anything in databases of protein structures determined experimentally or in the AlphaFold database of predictions from known organisms.
A good chunk of the AlphaFold database is made of structures that are nearly identical to each other, and ‘metagenomic’ databases “should cover a large part of the previously unseen protein universe”, says Martin Steinegger, a computational biologist at Seoul National University. “There’s a big opportunity now to unravel more of the darkness.”
Sergey Ovchinnikov, an evolutionary biologist at Harvard University in Cambridge, Massachusetts, wonders about the hundreds of millions of predictions that ESMFold made with low-confidence. Some might lack a defined structure, at least in isolation, whereas others might be non-coding DNA mistaken as a protein-coding material. “It seems there is still more than half of protein space we know nothing about,” he says.
A new way to explore topological superconductivity.
A new way to combine two materials with special electrical properties -- a monolayer superconductor and a topological insulator -- provides the best platform to date to explore an unusual form of superconductivity called topological superconductivity. The combination could provide the basis for topological quantum computers that are more stable than their traditional counterparts.
Superconductors -- used in powerful magnets, digital circuits, and imaging devices -- allow the electric current to pass without resistance, while topological insulators are thin films only a few atoms thick that restrict the movement of electrons to their edges, which can result in unique properties. A team led by researchers at Penn State describe how they have paired the two materials in a paper appearing Oct. 27th 2022 in the journal Nature Materials.
"The future of quantum computing depends on a kind of material that we call a topological superconductor, which can be formed by combining a topological insulator with a superconductor, but the actual process of combining these two materials is challenging," said Cui-Zu Chang, Henry W. Knerr Early Career Professor and Associate Professor of Physics at Penn State and leader of the research team. "In this study, we used a technique called molecular beam epitaxy to synthesize both topological insulator and superconductor films and create a two-dimensional heterostructure that is an excellent platform to explore the phenomenon of topological superconductivity."
In previous experiments to combine the two materials, the superconductivity in thin films usually disappears once a topological insulator layer is grown on top. Physicists have been able to add a topological insulator film onto a three-dimensional "bulk" superconductor and retain the properties of both materials. However, applications for topological superconductors, such as chips with low power consumption inside quantum computers or smartphones, would need to be two-dimensional.
In this paper, the research team stacked a topological insulator film made of bismuth selenide (Bi2Se3) with different thicknesses on a superconductor film made of monolayer niobium diselenide (NbSe2), resulting in a two-dimensional end-product. By synthesizing the heterostructures at very lower temperature, the team was able to retain both the topological and superconducting properties. "In superconductors, electrons form 'Cooper pairs' and can flow with zero resistance, but a strong magnetic field can break those pairs," said Hemian Yi, a postdoctoral scholar in the Chang Research Group at Penn State and the first author of the paper. "The monolayer superconductor film we used is known for its 'Ising-type superconductivity,' which means that the Cooper pairs are very robust against the in-plane magnetic fields. We would also expect the topological superconducting phase formed in our heterostructures to be robust in this way."
Combination of internal radiation and chemotherapy dissolves tumors in 80 percent of mice across multiple models
Biomedical engineers at Duke University have demonstrated the most effective treatment for pancreatic cancer ever recorded in mouse models. While most mouse trials consider simply halting growth a success, the new treatment completely eliminated tumors in 80% of mice across several model types, including those considered the most difficult to treat.
The approach combines traditional chemotherapy drugs with a new method for irradiating the tumor. Rather than delivering radiation from an external beam that travels through healthy tissue, the treatment implants radioactive iodine-131 directly into the tumor within a gel-like depot that protects healthy tissue and is absorbed by the body after the radiation fades away.
The results are published online October 19th 2022 in the journal Nature Biomedical Engineering. "We did a deep dive through over 1100 treatments across preclinical models and never found results where the tumors shrank away and disappeared like ours did," said Jeff Schaal. "When the rest of the literature is saying that what we're seeing doesn't happen, that's when we knew we had something extremely interesting." Despite accounting for only 3.2% of all cancer cases, pancreatic cancer is the third leading cause of cancer-related death. It is a very difficult to treat because its tumors tend to develop aggressive genetic mutations that make it resistant to many drugs, and it is typically diagnosed very late, when it has already spread to other sites in the body.
The current leading treatment combines chemotherapy, which keeps cells in a stage of reproduction vulnerable to radiation for longer periods of time, with a beam of radiation targeted at the tumor. This approach, however, is ineffective unless a certain threshold of radiation reaches the tumor. And despite recent advances in shaping and targeting radiation beams, that threshold is very difficult to reach without risking severe side effects.
Another method researchers have tried involves implanting a radioactive sample encased in titanium directly within the tumor. But because titanium blocks all radiation other than gamma rays, which travel far outside the tumor, it can only remain within the body for a short period of time before damage to surrounding tissue begins to defeat the purpose.
"There's just no good way to treat pancreatic cancer right now," said Schaal, who is now director of research at Cereius, Inc., a Durham, North Carolina biotechnology startup working to commercialize a targeted radionuclide therapy through a different technology scheme.
Researchers of the Wellcome Sanger Institute and ETH Zurich have created the first full connectivity map of the human immune system, showing how immune cells communicate with each other and ways to modulate these pathways in disease.
The immune system is made up of specialized cells, some of which individually travel through the body to scan for signs of injury or disease. Once these cells detect a threat, they need to communicate the message to other cells in order to mount an effective immune response. One way this cell-to-cell signaling is done is through proteins on the surfaces of cells that bind on to matching ‘receptor’ proteins on the surfaces of other cells. Previously, scientists and clinicians only had an incomplete map of these receptor connections between all of the different types of immune cells in the body.
Researchers from the Wellcome Sanger Institute (UK) and ETH Zurich are now filling these gaps. In a joint effort, they have created a first of its kind comprehensive map of the network of connections that make up the human immune system. In creating the immune system map, the scientists show how immune cells across the body link up and communicate. This research, published in Nature, includes the discovery of many previously unknown interactions that together shed light on the organization of the body’s immune defenses. An in-depth understanding of the interactions between immune cells, and how this communication fits into the human body as a whole, is vital if we are to develop treatments that enhance the immune system in order to fight disease, known as immunotherapies.
Immunotherapies have already demonstrated great potential in treating disease, most notably with certain cancers. However, these only work well in certain groups of patients and for particular conditions. Knowing the map of immune receptor connections could help explain why immunotherapies sometimes only work in a subset of patients, and offer new targets for designing future immunotherapies that may work for patients who currently do not benefit from these treatments.
A research group has systematically studied the properties of all heavy elements and decoded which ones formed during a recently discovered neutron star merger.
Details of this milestone were published in The Astrophysical Journal on October 26, 2022. When two neutron stars spiral inwards and merge, the resulting explosion produces a large amount of heavy elements that make up our Universe. The first confirmed example of this process was an event in 2017 named GW 170817. Yet, even five years later, identifying the specific elements created in neutron star mergers has eluded scientists, except for strontium identified in the optical spectra.
A research group led by Nanae Domoto, a graduate student at the Graduate School of Science at Tohoku University and a research fellow at the Japan Society for the Promotion of Science (JSPS), has systematically studied the properties of all heavy elements to decode the spectra from neutron star mergers.
They used this to investigate the spectra of kilonova - bright emissions caused by the radioactive decay of freshly synthesized nuclei that are ejected during the merger - from GW 170817. Based on comparisons of detailed kilonovae spectra simulations, produced by the supercomputer "ATERUI II" at the National Astronomical Observatory of Japan, the team found that the rare elements lanthanum and cerium can reproduce the near-infrared spectral features seen in 2017.
Until now, the existence of rare earth elements has only been hypothesized based on the overall evolution of the kilonova's brightness, but not from the spectral features. "This is the first direct identification of rare elements in the spectra of neutron star mergers, and it advances our understanding of the origin of elements in the Universe," Dotomo said. "This study used a simple model of ejected material. Looking ahead, we want to factor in multi-dimensional structures to grasp a bigger picture of what happens when stars collide," Dotomo added.
Pandemic stirs jitters about “gain-of-function” research
In a U.S. government lab in Bethesda, Maryland, virologists plan to equip the strain of the monkeypox virus that spread globally this year, causing mostly rash and flulike symptoms, with genes from a second monkeypox strain that causes more serious illness. Then they’ll see whether any of the changes make the virus more lethal to mice. The researchers hope that unraveling how specific genes make monkeypox more deadly will lead to better drugs and vaccines.
Some scientists are alarmed by the planned experiments, which were first reported by Science. If a more potent version of the outbreak strain accidentally escaped the high-containment, high-security lab at the National Institute of Allergy and Infectious Diseases (NIAID), it could spark an “epidemic with substantially more lethality,” fears epidemiologist Thomas Inglesby, director of the Center for Health Security at the Johns Hopkins University Bloomberg School of Public Health. That’s why he and others argue the experiments should undergo a special review required for especially risky U.S.-funded studies that might create a pathogen that could launch a catastrophic pandemic.
But it’s not clear that the rules apply to the proposed study. In a 2018, a safety panel determined it was exempt from review. Monkeypox did not meet the definition of a “potential pandemic pathogen” (PPP), the panel decided, because it didn’t spread easily. Now, with monkeypox widespread, the National Institutes of Health (NIH) is planning to reexamine the work, but it still might not qualify as “enhancing” a PPP, the agency says. That’s because the study will swap natural mutations, not create new ones, so it is not expected to create a monkeypox strain more virulent than the two already known.
The monkeypox controversy marks just the latest flare-up in a decade-old debate over exactly when a study that alters a pathogen is too risky for the U.S. government to fund—and who should have the power to decide. That wrangling became especially ferocious over the past 2 years, as the COVID-19 pandemic spawned allegations, so far unproven, that SARS-CoV-2 escaped from a laboratory in China. Now, in the pandemic’s wake, the U.S. government appears poised to make sizable changes to how it manages so-called gain-of-function (GOF) studies that tweak pathogens in ways that could make them spread faster or more dangerous to people.
Just recently, an expert panel convened by NIH and its parent agency, the Department of Health and Human Services (HHS), released a draft report that recommends the GOF rules be broadened to include pathogens and experiments that are exempt from the current scheme. If the recommendation is adopted—which could come next year—the monkeypox study could come under tighter scrutiny. And other researchers working with viruses such as Ebola, seasonal flu strains, measles, and even common cold viruses could face new oversight and restrictions.
Some scientists are watching nervously, worried that an expanded definition could worsen what they already see as a murky, problematic oversight system. The existing rules, they say, have caused confusion and delays that have deterred scientists from pursuing studies critical to understanding emerging pathogens and finding ways to fight them. If not implemented carefully, the proposed changes could “greatly impede research into evolving or emerging viruses,” worries virologist Linda Saif of Ohio State University, Wooster. She and others say expanding the regulations could add costly red tape, potentially driving research overseas or into the private sector, where U.S. regulations don’t apply or are looser.
Others say the proposed changes don’t go far enough. They’d like to see the U.S. government create an entirely new independent body to oversee risky research, and for the public to get far more information about proposed experiments that could have fearsome consequences. Some have even called for curbing the now common practice of collecting viruses from wild animals and studying them in the lab, saying it only increases the risks that the viruses—or modified versions—will jump to humans.
“We really should be asking important questions about whether that work should continue,” Inglesby says. And virologist James LeDuc, who retired last year as director of the University of Texas Medical Branch’s Galveston National Laboratory, says, “It’s one thing to recognize that these viruses exist in nature. It’s another to modify them so that you can study them if in fact they could become human pathogens.”
A team of researchers affiliated with several institutions in Canada and Denmark has found evidence that narwhals have been altering their seasonal migration patterns in response to global warming. In their paper published in Proceedings of the National Academy of Sciences, the group describes how they compared satellite data showing narwhal movements, with ice and temperature data in the Arctic over a 21-year period and what they learned by doing so.
Animals migrate in response to seasonal environments, to reproduce, to benefit from resource pulses, or to avoid fluctuating hazards. Although climate change is predicted to modify migration, only a few studies to date have demonstrated phenological shifts in marine mammals. In the Arctic, marine mammals are considered among the most sensitive to ongoing climate change due to their narrow habitat preferences and long life spans. Longevity may prove an obstacle for species to evolutionarily respond. For species that exhibit high site fidelity and strong associations with migration routes, adjusting the timing of migration is one of the few recourses available to respond to a changing climate. Now, scientists demonstrate evidence of significant delays in the timing of narwhal autumn migrations with satellite tracking data spanning 21 y from the Canadian Arctic. Measures of migration phenology varied annually and were explained by sex and climate drivers associated with ice conditions, suggesting that narwhals are adopting strategic migration tactics. Male narwhals were found to lead the migration out of the summering areas, while females, potentially with dependent young, departed later. Narwhals are remaining longer in their summer areas at a rate of 10 d per decade, a similar rate to that observed for climate-driven sea ice loss across the region. The consequences of altered space use and timing have yet to be evaluated but will expose individuals to increasing natural changes and anthropogenic activities on the summering areas.
MIT researchers have developed a technique for precisely controlling the arrangement and placement of nanoparticles on a material, like the silicon used for computer chips, in a way that does not damage or contaminate the surface of the material. The technique, which combines chemistry and directed assembly processes with conventional fabrication techniques, enables the efficient formation of high-resolution, nanoscale features integrated with nanoparticles for devices like sensors, lasers, and LEDs, which could boost their performance.
Transistors and other nanoscale devices are typically fabricated from the top down—materials are etched away to reach the desired arrangement of nanostructures. But creating the smallest nanostructures, which can enable the highest performance and new functionalities, requires expensive equipment and remains difficult to do at scale and with the desired resolution.
A more precise way to assemble nanoscale devices is from the bottom up. In one scheme, engineers have used chemistry to "grow" nanoparticles in solution, drop that solution onto a template, arrange the nanoparticles, and then transfer them to a surface. However, this technique also involves steep challenges.
First, thousands of nanoparticles must be arranged on the template efficiently. And transferring them to a surface typically requires a chemical glue, large pressure, or high temperatures, which could damage the surfaces and the resulting device.
The MIT researchers developed a new approach to overcome these limitations. They used the powerful forces that exist at the nanoscale to efficiently arrange particles in a desired pattern and then transfer them to a surface without any chemicals or high pressures, and at lower temperatures. Because the surface material remains pristine, these nanoscale structures can be incorporated into components for electronic and optical devices, where even minuscule imperfections can hamper performance.
"This approach allows you, through engineering of forces, to place the nanoparticles, despite their very small size, in deterministic arrangements with single-particle resolution and on diverse surfaces, to create libraries of nanoscale building blocks that can have very unique properties, whether it is their light-matter interactions, electronic properties, mechanical performance, etc.," says Farnaz Niroui, the EE Landsman Career Development Assistant Professor of Electrical Engineering and Computer Science (EECS) at MIT. "By integrating these building blocks with other nanostructures and materials we can then achieve devices with unique functionalities that would not be readily feasible to make if we were to use the conventional top-down fabrication strategies alone."
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...
