▲?作者：Sydney E. Towell, Maxim Ratushnyy et al.

▲?鏈接：https://www.nature.com/articles/s41586-025-08716-6

▲?摘要：

在此，我们通过C–H胺化与聚合物主链重排反应，读新将橡胶材料解构为环氧树脂前驱体。学网我们开发了一种硫二亚胺试剂，自然周论可对二烯类聚合物及橡胶实现高达约35%的出版烯丙位胺化修饰。随后利用阳离子2-aza-Cope重排反应，文导闻科成功解构胺化二烯聚合物。读新

在模型系统中，学网WhatsApp%E3%80%90+86%2015855158769%E3%80%91hoisting%20equipment我们看到聚合物分子量从58100 g mol?自然周论1降至约400 g mol?1；实际应用中，胺化处理的出版废旧橡胶经6小时反应即可转化为可溶性胺功能化聚合物，该产物可进一步制备具有与商用双酚A型树脂相当刚度的文导闻科环氧热固材料。这项工作证明了C–H胺化与主链重排使消费后材料的读新化学回收成为可能。

▲ Abstract：

Here we apply C–H amination and 学网backbone rearrangement of polymers to deconstruct these materials into precursors for epoxy resins. Specifically, we develop a sulfur diimide reagent that enables up to about 35% allylic amination of diene polymers and rubber. Then, we apply the cationic 2-aza-Cope rearrangement to deconstruct aminated diene polymers. In a model system, we see molecular weight reduction from 58,100 to approximately 400?g?mol?1, and aminated post-consumer rubber is deconstructed over 6?hours into soluble amine-functionalized polymers, which can be utilized to prepare epoxy thermosets with similar stiffnesses to commercial bisphenol A-derived resins. Altogether, this work demonstrates the power of C–H amination and backbone rearrangement to enable chemical recycling of post-consumer materials.

▲?作者：Sufi R. Ahmed, Reza Baghdadi et al.

▲?鏈接：https://www.nature.com/articles/s41586-025-08854-x

▲?摘要：

在此，我们介绍一款光子人工智能处理器，其可高效运行ResNet、BERT等先进AI模型，以及DeepMind最初展示的Atari深度强化学习算法。

该处理器在多数工作负载下达到接近电子计算精度的水平，标志着光子计算首次具备与成熟电子AI加速器竞争的实力，并为发展后晶体管计算技术迈出关键一步。

▲ Abstract：

Here we introduce a photonic AI processor that executes advanced AI models, including ResNet and BERT, along with the Atari deep reinforcement learning algorithm originally demonstrated by DeepMind. This processor achieves near-electronic precision for many workloads, marking a notable entry for photonic computing into competition with established electronic AI accelerators and an essential step towards developing post-transistor computing technologies.

▲?作者：Eslem Ben Arous, James A. Blinkhorn et al.

▲?鏈接：https://www.nature.com/articles/s41586-025-08613-y

▲?摘要：

在此，我们报道了中更新世晚期物质文化与科特迪瓦南部热带雨林环境的明确关联。

通过双生光激发发光和电子自旋共振测年方法，我们将Bété I遗址的人类活动年代精确限定在约15万年，并与智人直接关联。沉积物的植物蜡生物标志物、稳定同位素、植硅体及花粉分析均证实当时为湿润森林环境。

该成果呈现了迄今所知人类与热带雨林栖息地最早的确凿关联，其石制品组合与雨林环境的可靠对应关系表明：至迟在约15万年前，非洲森林已不构成智人扩散的显著生态屏障。

▲ Abstract：

Here we report a clear association between late Middle Pleistocene material culture and a wet tropical forest in southern C?te d’Ivoire, a region of present-day rainforest. Twinned optically stimulated luminescence and electron spin resonance dating methods constrain the onset of human occupations at Bété I to around 150?ka, linking them with Homo sapiens. Plant wax biomarker, stable isotope, phytolith and pollen analyses of associated sediments all point to a wet forest environment. The results represent the oldest yet known clear association between humans and this habitat type. The secure attribution of stone tool assemblages with the wet forest environment demonstrates that Africa’s forests were not a major ecological barrier for H. sapiens as early as around 150?ka.

▲?作者：Lin Cheng, Nan Wang et al.

▲?鏈接：https://www.nature.com/articles/s41586-024-08476-9

▲?摘要：

在此，构建了包含栽培二倍体马铃薯及其野生祖先种60个单倍型的相位化泛基因组图谱，并发现了转座元件在产生结构变异中的广泛存在。与线性参考基因组相比，该图谱呈现更广的遗传多样性（3076 Mb vs 742 Mb）。值得注意的是，栽培二倍体的杂合度显著高于野生种（14.0% vs 9.5%），表明马铃薯驯化过程中存在广泛杂交。

通过严格筛选，我们鉴定出19625个潜在的有害结构变异（dSVs），并揭示耦合相位中dSVs周围有害单核苷酸多态性（dSNPs）的偏向性累积。

基于此泛基因组图谱，我们通过计算设计出dSNPs和dSVs最小化的理想马铃薯单倍型。这些发现为无性繁殖的基因组基础提供了新见解，并将指导育开发优良自交系。

▲ Abstract：

Here we develop a phased potato pangenome graph of 60 haplotypes from cultivated diploids and the ancestral wild species, and find evidence for the prevalence of transposable elements in generating structural variants. Compared with the linear reference, the graph pangenome represents a broader diversity (3,076?Mb versus 742?Mb). Notably, we observe enhanced heterozygosity in cultivated diploids compared with wild ones (14.0% versus 9.5%), indicating extensive hybridization during potato domestication. Using conservative criteria, we identify 19,625 putatively deleterious structural variants (dSVs) and reveal a biased accumulation of deleterious single nucleotide polymorphisms (dSNPs) around dSVs in coupling phase. Based on the graph pangenome, we computationally design ideal potato haplotypes with minimal dSNPs and dSVs. These advances provide critical insights into the genomic basis of clonal propagation and will guide breeders to develop a suite of promising inbred lines.