Summary of Science publications, focusing on contents relevant to AI and life sciences
Proteins on the surface of cells are important disease markers and drug targets, but existing tools for finding aptamers—short nucleic acid sequences that bind specific proteins—that work directly on intact cells are inadequate for large-scale screening. Researchers developed SPARK-seq, a platform that combines single-cell RNA sequencing, aptamer sequencing, and CRISPR-based manipulation of surface protein expression to simultaneously characterize thousands of aptamer-protein interactions in one experiment. In a single run, the method mapped 5,535 distinct aptamers across eight surface proteins spanning a wide range of abundance levels. SPARK-seq could distinguish between closely related protein family members without cross-reactivity and extracted binding kinetics data, allowing researchers to identify aptamers that detach slowly from their targets—a desirable property for therapeutic use. These kinetic insights also guided the engineering of improved aptamer variants. The platform offers a systematic route toward discovering and optimizing aptamers for diagnostics and therapeutics.
Side-blotched lizards maintain three distinct male types — orange "usurpers," blue "mate-guarders," and yellow "sneakers" — that cycle in a rock-paper-scissors competitive dynamic. Researchers traced the genetic underpinning of this polymorphism, which had long been assumed to involve three alleles at a single locus. Genomic analysis revealed that orange and blue morphs are tied to two divergent haplotypes in the regulatory region of the sepiapterin reductase gene, while yellow sneakers appear to emerge through phenotypic plasticity from the same genetic background as blue morphs — not from a separate allele. Simulations showed that this two-allele-plus-plasticity architecture actually sustains the polymorphism more stably than a classic three-allele model would. The finding broadens understanding of how balancing selection operates, demonstrating that stable, cyclically competitive morphs in nature can be maintained through a hybrid system combining discrete genetic differences with developmental flexibility.
Somatic mutations that boost hematopoietic stem cell (HSC) fitness can drive clonal hematopoiesis (CH) and raise blood cancer risk, yet individuals vary considerably in how aggressively mutant clones expand. To explore whether inherited genetic differences underlie this variation, researchers identified a noncoding variant, rs17834140-T, that reduces expression of the RNA-binding protein MSI2 specifically in HSCs. Using computational modeling of variant effects and mapping of MSI2 binding targets, they characterized a posttranscriptional RNA regulatory network important for HSC maintenance. Carriers of the protective variant showed slower CH clone expansion, and MSI2 levels influenced how strongly ASXL1-mutant HSCs dominated the stem cell pool. The work demonstrates that naturally occurring genetic variation can confer resilience to CH by tuning RNA regulation, and points to MSI2 and its downstream targets as potential therapeutic entry points for preventing myeloid malignancies.
Aptamers are short nucleic acid sequences selected to bind specific targets, and their discovery traditionally relies on iterative enrichment cycles. This perspective examines how data science methods applied to aptamer selection libraries can uncover hidden patterns not apparent from conventional analysis. By analyzing the composition and sequence distributions within enriched pools, computational approaches reveal underlying structural and chemical features that correlate with binding ability. These insights could accelerate aptamer identification, reduce experimental burden, and improve understanding of what makes certain sequences effective binders. The work highlights an emerging intersection between machine learning or statistical analysis and nucleic acid selection, suggesting that treating aptamer libraries as rich datasets rather than simply collections of candidates opens new avenues for discovery and optimization.
Researchers examined how pigs were introduced across Wallacea and Oceania by combining genomic analysis of 576 pig nuclear genomes with geometric morphometric analysis of over 700 modern and ancient teeth. The study reveals that feral and domestic pigs across these island regions carry mixed ancestries, reflecting multiple successive introduction events accompanied by ongoing gene flow rather than a single founding population. Despite this genomic diversity, all these pigs share a distinctive tooth shape and a traceable genetic connection to Chinese domestic pig populations. This lineage arrived with the spread of Austronesian-speaking peoples roughly 3,000–4,000 years ago, moving through Taiwan and the Philippines. The findings clarify both the timing and routes of pig translocation across the Pacific and highlight how human migration and trade networks dramatically reshaped island ecosystems by introducing non-native animals over millennia.
Researchers developed a computational strategy for building protein complexes that assemble only in the presence of a specific small molecule. The approach exploits the symmetry of the ligand to guide design of matching protein interfaces. Using this method, they created protein homotrimers controlled by amantadine, an FDA-approved drug, and then extended the approach to heterodimers and heterotrimers also responsive to the same compound. X-ray crystal structures confirmed the designs were highly accurate, and biochemical tests validated that assembly depended on amantadine. The team then showed these engineered proteins could regulate where other proteins localize within cells, drive the formation of membraneless condensates, and switch gene expression on or off. Because natural small-molecule-regulated oligomerization systems are rare and hard to engineer, this work substantially expands the chemogenetic toolkit available for precisely controlling cellular processes with drug-like small molecules.
Research into how cooperation emerges in both animals and AI systems has revealed striking similarities between the two. Scientists trained mice and artificial agents on tasks requiring coordinated action to earn shared rewards, then compared the underlying mechanisms. In mice, specific social behavioral strategies and decision-making processes drove successful cooperation, and these were represented in the anterior cingulate cortex (ACC) — a brain region whose activity was shown through causal manipulation to directly support cooperative behavior. When artificial agents were trained on an analogous task, they independently developed behavioral strategies and internal representations that closely mirrored those found in biological brains. The findings suggest that cooperative intelligence, whether biological or artificial, may converge on similar computational solutions, and they highlight the ACC as a key neural substrate for social coordination. This work could inform the design of AI systems better suited to collaborative environments.
Comments (0)
No comments yet. Log in to leave a comment.
You must be logged in to leave a comment.
Login to Comment