Data Science

Understanding diseases and designing treatments for them requires knowing what proteins are involved and their precise shapes. But the knotty twists and folds of most proteins have remained elusive, until now. In July, London AI lab DeepMind released the structure of every protein in the human body, along with 20 research animal models. The data set—and the AI program, AlphaFold, that created it—are available for free to researchers. The company worked with EMBL’s European Bioinformatics Institute to host a searchable database.

Capitalizing on the downtime of pandemic lockdowns, a University of Reading climate scientist put out a call in March 2020 for help filling in historical rainfall data to better understand climatic variations in the region. The Rainfall Rescue Project received around 16,000 volunteers’ information from more than 65,000 handwritten rainfall records dating back to the 1600s. In a year, the volunteer team had created a digital database of over 5 million observations, helping water companies better understand and plan for wet and dry periods.

The largest expedition of its kind, the MOSAiC research mission brought a total of 442 researchers, crewmembers and others from more than three dozen countries to the northernmost parts of the planet to collect an unprecedented trove of atmospheric, marine and ice data. Centered around Polarstern, a massive icebreaker ship, the expedition marked the first time scientists were able to study ice and weather in the region in any climate and season, including frigid, dark winters. Led by Germany’s Alfred Wegener Institute, the €150 million collaboration is designed to help scientists draw better global climate models.

Over three years, the Elephant Listening Project eavesdropped on the rainforest of the Republic of Congo via 50 microphones hidden in trees. The mission was to listen for the call of elusive forest elephants to understand where they are, what they’re doing, and what they’re trying to communicate. But the recordings also captured hundreds—possibly thousands—of other species. In June, the project released all 1 million hours of soundscapes for researchers all over the world to use in their own biodiversity efforts.

Tracking deforestation in the Amazon is crucial to fighting it, but it’s difficult to do. U.S. tech giant SAS teamed up with Austria’s International Institute for Applied Systems Analysis to build an AI algorithm that can spot signs of deforestation from satellite images. But because computers sometimes miss clues humans could catch, the partners launched a public web app that lets citizen scientists identify forest loss—in the process, teaching the computers what to look for. The first phase classified human impacts in the Amazon from 90,000 satellite images.

In the roughly two decades since the Human Genome Project sequenced the human genome, scientists have been trying to chip away at the missing parts. Yet some 8 percent of our genetic coding still remained a mystery. In May, the Telomere-to-Telomere Consortium (T2T), made up of more than two dozen international institutions, announced it had filled in almost all the gaps, adding 200 million new base pairs—comprising 115 new genes—to the known genome.

Working with NASA, the European Union’s Copernicus program and other institutions and government agencies, Google added historical data to Google Earth to visualize the global impacts of development and climate change over time. The Timelapse feature has been around since 2013, but in April, the company released its biggest update in five years. The new version includes 24 million satellite pictures taken between 1984 and 2020—20 million gigabytes of imagery—so users can see how specific places have changed over time. One of the goals? Let researchers use Timelapse to understand and combat destruction such as wildfires and deforestation to avoid future disasters.

Imagine a team of aerial drones and terrestrial bots put to use in search and rescue missions, looking for lost hikers or survivors of a disaster. Super promising—but there had to be a way to get these techno-teams working together efficiently and not wasting energy on a convoluted search trajectory. And researchers at the Massachusetts Institute of Technology found it: distributed local search. A collaboration algorithm to ensure maximum information gathering with minimal wasted energy, the iterative approach improves the team’s performance by adding or removing individual robots’ trajectories from the group’s overall plan.

What do whales say? That’s the central question of Project CETI, the Cetacean Translation Initiative, which will use underwater microphones, AI and cryptography to try to translate the calls of sperm whales in the Caribbean Sea while attempting to communicate back. The project’s main goal is to show how cutting-edge technologies can benefit not only humans but other species as well. Launched in April 2021, the five-year project is funded by The Audacious Project, a collaborative funding initiative housed at TED.

With the deluge of digital ads comes a surge in misleading and scammy advertisements. To combat the trend, the U.K.’s Advertising Standards Authority developed a new technology system capable of identifying and flagging false claims and fraud. Working with Instagram, for instance, the group managed to remove digital ads for organic Botox, with 12,000 reports sent to the platform. The effort is part of an upcoming Online Platforms and Network Standards project.