Lluís Montoliu

In 2025, a six-year-old Chinese girl named Mei (not her real name) became the first person in the world to undergo gene-editing therapy targeting the brain. She had a developmental delay associated with a rare genetic syndrome, Snijders-Blok-Campeau syndrome. According to Science, her parents had raised $860,000 to partially fund the treatment at Xinhua Hospital in Shanghai (China), led by Zilong Qiu, one of the neuroscientists competing to apply base editors — a version of CRISPR — in children with rare diseases. Now, Science and Retraction Watch have exclusively revealed that Mei died from a severe immune reaction seven days after receiving the treatment, a story that had been kept hidden. The entry on clinicaltrials.gov has not been updated since 2025; and when Qiu and his team published animal studies related to the trial in the journal Nature in early 2026, according to the journalistic investigation, they omitted any reference to Mei and her family.

The range of CRISPR tools for genome editing can extend beyond nature-inspired designs thanks to proteins designed using artificial intelligence. A new study by Nobel laureate Jennifer Doudna’s team, published in Science, describes the design of synthetic RNA-guided nucleases, with sequences substantially different from those found in nature, which match or exceed the activity of their natural counterparts whilst offering novel properties.

An article published in Nature describes how the first use of precision editing has shed light on a gene essential for embryonic development. The authors caution that the clinical application of genome editing in human embryos requires rigorous ethical analysis and oversight, as well as broad public debate and support.

We often see headlines claiming that new research has found a ‘cure’ for diseases. However, what are the real chances of this being true? How does the current stage of the research affect its ultimate outcome? Has it already been tested on humans, or only on laboratory animals? In this article, we explain why animals are used in biomedicine, the reasons why mice are the most commonly used, the steps and timeframes involved from when a treatment appears effective until it can be determined whether it works in people, the characteristics and limitations of various disease models, and how results should be communicated to inform the public without raising false hopes.

The US company Colossal Biosciences claims to have successfully incubated a bird in an artificial egg system until it hatched. In a press release, the company states that this technology allows a bird embryo to develop fully outside the shell of a biological egg, and could be used to bring back extinct bird species such as the giant moa from New Zealand’s South Island.

A team from the United States analyzed 611 samples from 341 model mouse strains stored at the Mutant Mouse Resource and Research Centers (MMRRC), a research resource network supported by the National Institutes of Health (NIH). By comparing the identity of each strain with its actual genetic profile, they found that approximately half of the samples showed discrepancies. Although the expected engineered mutation was generally present and many inconsistencies were relatively minor, some had the potential to compromise the validity and reproducibility of the experiments by introducing hidden genetic variables that could alter biological outcomes. The findings are published in Science.

Craig Venter, the American biologist and entrepreneur who founded Celera Genomics to launch his own Human Genome Project in 1999 outside the public consortium, died Wednesday in San Diego at the age of 79, according to a statement from the J. Craig Venter Institute, which he led. Among other achievements, Venter completed the first full sequencing of a living organism’s genetic material and announced that he had succeeded in creating synthetic life.

Repeated cloning cannot be sustained indefinitely in mammals, according to the findings of a twenty-year study on mice conducted in Japan. Serial cloning of mice led to an accumulation of lethal DNA mutations that affected birth rates from the 27th generation onwards, with the 58th generation being the last, according to the article published in Nature Communications, showing that sexual reproduction is necessary to prevent large-scale genetic mutations.

AlphaGenome is a deep learning model developed by Google DeepMind capable of predicting the function of DNA sequences up to one million base pairs long. An evaluation of the tool shows that it matches or improves upon the predictive ability of existing models in 25 of the 26 tests performed. According to the authors, who are part of Google DeepMind itself, AlphaGenome can help scientists "better understand genome function, the biology of diseases, and ultimately drive new biological discoveries and the development of new treatments." The results are published in Nature. 

Two studies published today in the journal Nature significantly expand the catalogue of known human genetic variations. The resulting data constitute what may be the most complete view of the human genome to date.