Hello Nature readers, would you like to get this Briefing in your inbox free every day? Sign up here.
A pair of Tasmanian tigers photographed at an Australian zoo in 1933.Credit: Universal History Archive/Universal Images Group via Getty
Tasmanian tiger RNA recovered
For the first time, researchers have sequenced RNA from an extinct species — the thylacine — using samples from a 132-year-old museum specimen. Because RNA rapidly breaks down outside living cells, the sequences were pieced together from millions of smaller fragments. Being able to look at RNA “opens up a whole new potential source of information”, says geneticist Oliver Smith. “As opposed to looking at what a genome is, we can look at what the genome does.”
Reference: Genome Research paper
Powerful laser will ‘film’ chemistry
The world’s most powerful X-ray laser will make it possible to create super-sharp movies of ultrafast processes, such as electrical charges hopping around atoms during a chemical reaction. A US$1.1-billion upgrade to the Linac Coherent Light Source (LCLS) in California increased the laser’s pulse firing rate roughly 8,000-fold and its brightness by, on average, 10,000-fold. “We are waiting for the LCLS-II so we can do our dream experiment that we’ve been preparing for ten years,” says molecular biophysicist Junko Yano.
AlphaFold tool pinpoints disease mutations
An artificial intelligence (AI) network called AlphaMissense can predict which mutations in proteins are likely to cause health conditions. The tool is based on Google DeepMind’s AlphaFold, which predicts protein structures from their amino-acid sequence. AlphaMissense seems to outperform similar tools but is “not a gigantic leap forward”, says computational biologist Arne Elofsson.
NIH upholds controversial plan
The US National Institutes of Health (NIH) will move forward with its plan that foreign scientists must share laboratory notebooks and raw data with their US research partners. The policy is directed at holders of ‘subawards’’ — funds that a US-based researcher can give to a collaborator — and arose because of controversy over how the NIH oversaw subawards to the Wuhan Institute of Virology in China. Scientists greeted the initial plan with criticism, so the NIH has changed the mandated data-sharing frequency from every few months to once per year and delayed implementation until March 2024. The revisions are “cosmetic changes”, says physician-scientist Gerald Keusch. The policy remains too broad and suggests a lack of trust in foreign researchers, he adds.
Features & opinion
A biological delivery service for your cells
Extracellular vesicles (EVs), minuscule, cargo-carrying bubbles, could one day deliver gene therapies or target cancer cells for destruction. Cells use EVs to remove waste, communicate and move molecules between cells — a mechanism researchers want to exploit for diagnosing or treating diseases. The number of patent applications and clinical trials has soared. In parallel, a shadier market has developed: hundreds of private clinics and companies sell unproven EV-based therapies, offering to treat anything from Alzheimer’s disease to baldness.
Why we must bring back samples from Mars
On Sunday, if all goes to plan, a tiny capsule of material from an asteroid will be dropped in the desert in Utah. The sample travelled two billion kilometres in the grasp of the OSIRIS-REx spacecraft, which snatched it during a brief touchdown two years ago. The exhilarating moment should whet the appetite for a similar sample-return mission from Mars, argues Meenakshi Wadhwa, the principal scientist for just such a mission at NASA. “Sample-return missions have too few cheerleaders, even among planetary scientists,” she writes. Yes, it will be expensive, argues Wadhwa — but the scientific gains will be worth it.
For drug discovery, AI needs data
Researchers must find a way to train algorithms collaboratively without revealing competitive information, argues a group of drug-discovery scientists. AI has the potential to speed up drug development, but machine-learning models need more training data than a single company’s trials can provide. The group encourages biopharmaceutical companies to consider federated and active learning. In the former, separate teams train a model without sharing the underlying data, while the latter helps to find data gaps that need to be filled.
Source: M. Mock et al.
