Every new paper, in one sentence you can actually read.

Each issue takes 10 brand-new scholarly papers, says in a single sentence what each one found, then explains it. Every paper is screened for retractions and a live DOI before it reaches you.

This screen checks process signals. It does not tell you the findings are true.

Get it free, every two days

Five papers free every issue, about 15 emails a month. No spam. Leave in one click.

Prefer no email? The same issues are in the RSS feed.

In this issue 10 papers · 5 free

01 Science A statistical test for the benefits of personalizing interventions Free
02 Science Roots navigate around decay regions by sensing local pH gradients Free
03 Nature Communications Bioinspired microcapsule reactor with engineered probiotics for IBD therapy Free
04 Nature Medicine Health system learning enables generalist neuroimaging models Free
05 Nature Architecture of the 8 MDa Hdr-Vhu-Fwd super-assembly in class I methanogens Free
06 Nature Ancient feeding-related neuropeptides regulate alloparenting in ants Subscribers
07 Nature Communications Atmospheric CO₂-to-acetaldehyde artificial photosynthesis in metallo hydrogen-bonded organic frameworks Subscribers
08 Proceedings of the National Academy of Sciences Localized sample-based quantum diagonalization for strongly correlated chemistry Subscribers
09 Proceedings of the National Academy of Sciences Multiplexing in networks and diffusion Subscribers
10 Nature Human Behaviour Common and rare genetic variant associations with cognitive performance across development in British birth cohorts Subscribers

The papers Read free

Science·9 July 2026

Zhaoqi Li, Emma Brunskill

A statistical test for the benefits of personalizing interventions

A new tool checks if giving people different, custom treatments works better than giving everyone the same single best option. It uses past data to make this check. The test keeps a low rate of false alarms. It was tried on data from job training, depression care, schools, and recommendation systems. In these tests, it worked well and beat other methods.

A new statistical test can show if custom treatments beat one-size-fits-all options.

The number: No key statistic identified
The caveat: This is a math tool, not a treatment itself. It does not test any single program directly on real people. It only checks old data, so it cannot prove personalizing will always work in new situations.

For your life: today, nothing. This is basic research.

Integrity screen: passed (3 checks) Checked 15 July 2026. Retraction record: none. DOI resolves at doi.org. Metadata record found (Science). Read the source

Science·9 July 2026

Zhulatai Bao, Huihui Wang et al.

Roots navigate around decay regions by sensing local pH gradients

Plant roots can bend away from rotting plant material in soil, without ever touching it. Rotting plant matter is broken down by fungi and other microbes. This breakdown releases acids that spread through the soil and change how acidic the ground is nearby. Root cells sense this uneven acid pattern and use it to grow away from the decaying spot. The root does this by turning the chemical signal into a change inside its own cells that makes one side grow differently than the other, so the root curves away.

Roots can detect and steer away from rotting plant matter using acid signals in soil.

The number: No key statistic identified
The caveat: The abstract gives no sample sizes, plant types tested, or bending amounts. So it is hard to know how strong or common this effect is. This is basic plant biology research. It does not tell us how this might affect farming or crops grown in real fields.

For your life: today, nothing. This is basic research.

Integrity screen: passed (3 checks) Checked 15 July 2026. Retraction record: none. DOI resolves at doi.org. Metadata record found (Science). Read the source

Nature Communications·13 July 2026

Minhui Xu, Yuanchun Du et al.

Bioinspired microcapsule reactor with engineered probiotics for IBD therapy

Scientists built a tiny capsule that carries special bacteria safely through the stomach to the gut. The capsule protects the bacteria from stomach acid. Once the bacteria reach inflamed spots in the intestine, they sense the problem. Then they release a calming substance that fights inflammation. In mice with bowel disease, the treatment repaired the gut lining. It also lowered inflammation and cell stress. It helped restore healthy gut bacteria too. The effect worked by blocking a specific inflammation signal inside cells.

A protective capsule delivered gut bacteria that healed inflamed intestines in mice.

The number: No key statistic identified
The caveat: This was tested only in male mice, not in people. So the results may not apply to humans yet.

For your life: nothing to act on yet. Early research, worth watching.

Integrity screen: passed (3 checks) Checked 13 July 2026. Retraction record: none. DOI resolves at doi.org. Metadata record found (Nature Communications). Read the source

Nature Medicine·10 July 2026

Akhil Kondepudi, Akshay Rao et al.

Health system learning enables generalist neuroimaging models

A computer model learned to read brain scans by studying millions of real hospital images. It now beats today's top general AI systems at this task. Its makers call this method "health system learning." General AI systems trained on public internet data do poorly with brain scans. This happens because privacy rules keep most medical scans off the public internet. The new model, called NeuroVFM, was trained on over 5 million brain scans from real patient care. It can spot disease patterns and write radiology reports. When paired with a separate tool that writes text, it produced reports that were more accurate. Experts also preferred these reports over ones from leading general AI systems. The model also made fewer false or missed findings.

A brain scan AI trained on real hospital data beat top general AI systems at reading scans.

The number: The model was trained on 5.24 million clinical brain scans from routine hospital care.
The caveat: This is a lab and technical study, not a real world clinical trial. It cannot tell us how the tool would perform if used daily by doctors on new patients.

For your life: nothing to act on yet. Early research, worth watching.

Integrity screen: passed (3 checks) Checked 23 August 2026. Retraction record: none. DOI resolves at doi.org. Metadata record found (Nature Medicine). Read the source

Nature·8 July 2026

Sophia Paul, Tomas C. Pascoa et al.

Architecture of the 8 MDa Hdr-Vhu-Fwd super-assembly in class I methanogens

Tiny microbes called methanogens make most of the world's methane, a strong greenhouse gas. Scientists mapped the detailed shape of a huge molecular machine these microbes use to make energy. The machine has 252 protein chains and more than 600 tiny parts that move electrons. It forms two ring shaped sections joined by a middle piece. Together they make one big loop that moves electrons in a circle. This loop connects the last step and the first step of methane production. The machine can also swap one part for a different one. This lets the microbe adjust to different surroundings. This exact design only shows up in one major group of methanogens, not in others.

A giant 8 million dalton machine in methane making microbes forms one continuous electron loop.

The number: The complex contains 252 protein chains and more than 600 redox cofactors, weighing about 8 megadaltons.
The caveat: This work shows the machine's shape and parts. It cannot fully prove how each step works inside a living cell.

For your life: today, nothing. This is basic research.

Integrity screen: passed (3 checks) Checked 22 July 2026. Retraction record: none. DOI resolves at doi.org. Metadata record found (Nature). Read the source

5 more papers are in the full issue.

The same screening, the full reading, every two days. 2 euros a month, or 20 a year, and you can leave in one click.

Nature·8 July 2026

Alexander Paul, Tomas Kay et al.