Frederic Tremblay, Qiang Xiong, Shrijal S. Shah, Chih-Wei Ko, Kenneth Kelly, Mary S. Morrison, Cristiana Giancarlo, Ricardo N. Ramirez, Erica M. Hildebrand, Sarah B. Voytek, Gabriel K. El Sebae, Shane H. Wright, Liam Lofgren, Scott Clarkson, Christine Waters, Samantha J. Linder, Songlei Liu, Taesun Eom, Shefal Parikh, Yuki Weber, Salette Martinez, Padma Malyala, Sahar Abubucker, Ari E. Friedland, Morgan L. Maeder, Angelo Lombardo, Vic E. Myer, Aron B. Jaffe <<<

Large Language Models (LLMs) have demonstrated strong capabilities but remain fundamentally static, unable to adapt their internal parameters to novel tasks, evolving knowledge domains, or dynamic interaction contexts. As LLMs are increasingly deployed in open-ended, interactive environments, this static nature has become a critical bottleneck, necessitating agents that can adaptively reason, act, and evolve in real time. This paradigm shift -- from scaling static models to developing self-evolving agents -- has sparked growing interest in architectures and methods enabling continual learning and adaptation from data, interactions, and experiences. This survey provides the first systematic and comprehensive review of self-evolving agents, organized around three foundational dimensions -- what to evolve, when to evolve, and how to evolve. We examine evolutionary mechanisms across agent components (e.g., models, memory, tools, architecture), categorize adaptation methods by stages (e.g., intra-test-time, inter-test-time), and analyze the algorithmic and architectural designs that guide evolutionary adaptation (e.g., scalar rewards, textual feedback, single-agent and multi-agent systems). Additionally, we analyze evaluation metrics and benchmarks tailored for self-evolving agents, highlight applications in domains such as coding, education, and healthcare, and identify critical challenges and research directions in safety, scalability, and co-evolutionary dynamics. By providing a structured framework for understanding and designing self-evolving agents, this survey establishes a roadmap for advancing adaptive agentic systems in both research and real-world deployments, ultimately shedding lights to pave the way for the realization of Artificial Super Intelligence (ASI), where agents evolve autonomously, performing at or beyond human-level intelligence across a wide array of tasks. <<<

Jiale Xiang, Xiangzhong Sun, Jiguang Peng, Hongfu Zhang, Jiankun Shen, Jingrou Li, Hongyu Li, Lanping Hu, Jingjing Zhang, Shihao Zhou, Sihu Xu, Yun Yang, Jun He, Zhiyu Peng <<<

Jiaqi You, Zhenping Liu, Zhengyang Qi, Yizan Ma, Mengling Sun, Ling Su, Hao Niu, Yabing Peng, Xuanxuan Luo, Mengmeng Zhu, Yuefan Huang, Xing Chang, Xiubao Hu, Yuqi Zhang, Ruizhen Pi, Yuqi Liu, Qingying Meng, Jianying Li, Qinghua Zhang, Longfu Zhu, Zhongxu Lin, Ling Min, Daojun Yuan, Corrinne E. Grover, David D. Fang, Keith Lindsey, Jonathan F. Wendel, Lili Tu, Xianlong Zhang, Maojun Wang <<<

AbstractPolyploidy complicates transcriptional regulation and increases phenotypic diversity in organisms. The dynamics of genetic regulation of gene expression between coresident subgenomes in polyploids remains to be understood. Here we document the genetic regulation of fiber development in allotetraploid cotton Gossypium hirsutum by sequencing 376 genomes and 2,215 time-series transcriptomes. We characterize 1,258 genes comprising 36 genetic modules that control staged fiber development and uncover genetic components governing their partitioned expression relative to subgenomic duplicated genes (homoeologs). Only about 30% of fiber quality-related homoeologs show phenotypically favorable allele aggregation in cultivars, highlighting the potential for subgenome additivity in fiber improvement. We envision a genome-enabled breeding strategy, with particular attention to 48 favorable alleles related to fiber phenotypes that have been subjected to purifying selection during domestication. Our work delineates the dynamics of gene regulation during fiber development and highlights the potential of subgenomic coordination underpinning phenotypes in polyploid plants. <<<

The Mycobacterium tuberculosis complex (MTBC) is a group of bacteria causing tuberculosis (TB) in humans and animals. Understanding MTBC genetic diversity is crucial for insights into its adaptation and traits related to survival, virulence, and antibiotic resistance. While it is known that within-MTBC diversity is characterised by large deletions found only in certain lineages (regions of difference [RDs]), a comprehensive pangenomic analysis incorporating both coding and non-coding regions remains unexplored. We utilised a curated dataset representing various MTBC genomes, including under-represented lineages, to quantify the full diversity of the MTBC pangenome. The MTBC was found to have a small, closed pangenome with distinct genomic features and RDs both between lineages (as previously known) and between sub-lineages. The accessory genome was identified to be a product of genome reduction, showing both divergent and convergent deletions. This variation has implications for traits like virulence, drug resistance, and metabolism. The study provides a comprehensive understanding of the MTBC pangenome, highlighting the importance of genome reduction in its evolution, and underlines the significance of genomic variations in determining the pathogenic traits of different MTBC lineages. <<<

This paper introduces Genesis, the first compiler designed to supportHamiltonian Simulation on hybrid continuous-variable (CV) and discrete-variable(DV) quantum computing systems. Genesis is a two-level compilation system. Atthe first level, it decomposes an input Hamiltonian into basis gates using thenative instruction set of the target hybrid CV-DV quantum computer. At thesecond level, it tackles the mapping and routing of qumodes/qubits to implementlong-range interactions for the gates decomposed from the first level. Ratherthan a typical implementation that relies on SWAP primitives similar toqubit-based (or DV-only) systems, we propose an integrated design ofconnectivity-aware gate synthesis and beamsplitter SWAP insertion tailored forhybrid CV-DV systems. We also introduce an OpenQASM-like domain-specificlanguage (DSL) named CVDV-QASM to represent Hamiltonian in terms ofPauli-exponentials and basic gate sequences from the hybrid CV-DV gate set.Genesis has successfully compiled several important Hamiltonians, including theBose-Hubbard model, $\mathbb{Z}_2-$Higgs model, Hubbard-Holstein model,Heisenberg model and Electron-vibration coupling Hamiltonians, which arecritical in domains like quantum field theory, condensed matter physics, andquantum chemistry. Our implementation is available atGenesis-CVDV-Compiler(https://github.com/ruadapt/Genesis-CVDV-Compiler). <<<

Akanksha Jain, Gilles Gut, Fátima Sanchis-Calleja, Reto Tschannen, Zhisong He, Nicolas Luginbühl, Fides Zenk, Antonius Chrisnandy, Simon Streib, Christoph Harmel, Ryoko Okamoto, Malgorzata Santel, Makiko Seimiya, René Holtackers, Juliane K. Rohland, Sophie Martina Johanna Jansen, Matthias P. Lutolf, J. Gray Camp, Barbara Treutlein <<<

Abstract Brain organoids enable the mechanistic study of human brain development and provide opportunities to explore self-organization in unconstrained developmental systems1–3. Here we establish long-term, live light-sheet microscopy on unguided brain organoids generated from fluorescently labelled human induced pluripotent stem cells, which enables tracking of tissue morphology, cell behaviours and subcellular features over weeks of organoid development4. We provide a novel dual-channel, multi-mosaic and multi-protein labelling strategy combined with a computational demultiplexing approach to enable simultaneous quantification of distinct subcellular features during organoid development. We track actin, tubulin, plasma membrane, nucleus and nuclear envelope dynamics, and quantify cell morphometric and alignment changes during tissue-state transitions including neuroepithelial induction, maturation, lumenization and brain regionalization. On the basis of imaging and single-cell transcriptome modalities, we find that lumenal expansion and cell morphotype composition within the developing neuroepithelium are associated with modulation of gene expression programs involving extracellular matrix pathway regulators and mechanosensing. We show that an extrinsically provided matrix enhances lumen expansion as well as telencephalon formation, and unguided organoids grown in the absence of an extrinsic matrix have altered morphologies with increased neural crest and caudalized tissue identity. Matrix-induced regional guidance and lumen morphogenesis are linked to the WNT and Hippo (YAP1) signalling pathways, including spatially restricted induction of the WNT ligand secretion mediator (WLS) that marks the earliest emergence of non-telencephalic brain regions. Together, our work provides an inroad into studying human brain morphodynamics and supports a view that matrix-linked mechanosensing dynamics have a central role during brain regionalization. <<<

Wen-Yan Cheng, Jian-Feng Li, Yong-Mei Zhu, Xiang-Jie Lin, Li-Jun Wen, Fan Zhang, Yu-Liang Zhang, Ming Zhao, Hai Fang, Sheng-Yue Wang, Xiao-Jing Lin, Niu Qiao, Wei Yin, Jia-Nan Zhang, Yu-Ting Dai, Lu Jiang, Xiao-Jian Sun, Yi Xu, Tong-Tong Zhang, Su-Ning Chen, Hong-Hu Zhu, Zhu Chen, Jie Jin, De-Pei Wu, Yang Shen, Sai-Juan Chen <<<

The current classification of acute myeloid leukemia (AML) relies largely on genomic alterations. Robust identification of clinically and biologically relevant molecular subtypes from nongenomic high-throughput sequencing data remains challenging. We established the largest multicenter AML cohort (n = 655) in China, with all patients subjected to RNA sequencing (RNA-Seq) and 619 (94.5%) to targeted or whole-exome sequencing (TES/WES). Based on an enhanced consensus clustering, eight stable gene expression subgroups (G1–G8) with unique clinical and biological significance were identified, including two unreported (G5 and G8) and three redefined ones (G4, G6, and G7). Apart from four well-known low-risk subgroups including PML::RARA (G1), CBFB::MYH11 (G2), RUNX1::RUNX1T1 (G3), biallelic CEBPA mutations or -like (G4), four meta-subgroups with poor outcomes were recognized. The G5 (myelodysplasia-related/-like) subgroup enriched clinical, cytogenetic and genetic features mimicking secondary AML, and hotspot mutations of IKZF1 (p.N159S) (n = 7). In contrast, most NPM1 mutations and KMT2A and NUP98 fusions clustered into G6–G8, showing high expression of HOXA / B genes and diverse differentiation stages, from hematopoietic stem/progenitor cell down to monocyte, namely HOX -primitive (G7) , HOX -mixed (G8), and HOX -committed (G6). Through constructing prediction models, the eight gene expression subgroups could be reproduced in the Cancer Genome Atlas (TCGA) and Beat AML cohorts. Each subgroup was associated with distinct prognosis and drug sensitivities, supporting the clinical applicability of this transcriptome-based classification of AML. These molecular subgroups illuminate the complex molecular network of AML, which may promote systematic studies of disease pathogenesis and foster the screening of targeted agents based on omics. <<<

Woody Ahern, Jason Yim, Doug Tischer, Saman Salike, Seth M. Woodbury, Donghyo Kim, Indrek Kalvet, Yakov Kipnis, Brian Coventry, Han Raut Altae-Tran, Magnus Bauer, Regina Barzilay, Tommi S. Jaakkola, Rohith Krishna, David Baker <<<

AbstractDe novoenzyme design starts from ideal active site descriptions consisting of constellations of catalytic residue functional groups around reaction transition state(s), and seeks to generate protein structures that can accurately hold the site in place. Highly active enzymes have been designed starting from such descriptions using the generative AI method RFdiffusion [1–3], but there are two current methodological limitations. First, the geometry of the active site can only be specified at the residue level, so for each catalytic residue functional group placed around the reaction transition state, the possible locations of the residue backbone must be enumerated by building side chain rotamers back from the functional group. Second, the location of the catalytic residues along the sequence must be specified in advance, which considerably limits the space of solutions which can be sampled. Here we describe a new deep generative method, Rosetta Fold diffusion 2 (RFdiffusion2), that solves both problems, enabling enzymes to be designed from sequence agnostic descriptions of functional group locations without inverse rotamer generation. We first evaluate RFdiffusion2 on anin silicoenzyme design benchmark of 41 diverse active sites and find that it is able to successfully build proteins scaffolding all 41 sites, compared to 16/41 with prior state-of-the-art deep learning methods. Next, we design enzymes around three diverse catalytic sites and characterize the designs experimentally; in each case we identify active catalysts in testing less than 96 sequences. RFdiffusion2 demonstrates the potential of atomic resolution generative models for the design ofde novoenzymes directly from their reaction mechanisms. <<<

Gayani Senevirathne, Serena C. Fernandopulle, Daniel Richard, Stephanie L. Baumgart, Anika Liv Christensen, Matteo Fabbri, Jakob Höppner, Harald Jüppner, Peishu Li, Vivien Bothe, Nadia Fröbisch, Ian Simcock, Owen J. Arthurs, Alistair Calder, Naomi Freilich, Niamh C. Nowlan, Ian A. Glass, April Craft, Terence D. Capellini <<<

Emily Grist, Peter Dutey-Magni, Marina A. Parry, Larissa Mendes, Ashwin Sachdeva, James A. Proudfoot, Anis A. Hamid, Mazlina Ismail, Sarah Howlett, Stefanie Friedrich, Lia DePaula Oliveira, Laura Murphy, Christopher Brawley, Oluwademilade Dairo, Sharanpreet Lall, Yang Liu, Daniel Wetterskog, Anna Wingate, Karolina Nowakowska, Leila Zakka, Claire L. Amos, Nafisah B. Atako, Victoria Wang, Hannah L. Rush, Robert J. Jones, Hing Leung, William R. Cross, Silke Gillessen, Chris C. Parker, Teresa Marafioti, Alfonso Urbanucci, Matthew Fittall, Edward M. Schaeffer, Daniel E. Spratt, David Waugh, Thomas Powles, Matthew R. Sydes, Felix Y. Feng, Daniel M. Berney, Mahesh K.B. Parmar, Noel W. Clarke, Elai Davicioni, Tamara L. Lotan, Christopher J. Sweeney, Louise C. Brown, Nicholas D. James, Gerhardt Attard <<<

Abstract Motivation The 3D organization of the genome plays a crucial role in various biological processes. Hi-C technology is widely used to investigate chromosome structures by quantifying 3D proximity between genomic regions. While numerous computational tools exist for detecting differences in Hi-C data between conditions, a comprehensive review and benchmark comparing their effectiveness is lacking. Results This study offers a comprehensive review and benchmark of 10 generic tools for differential analysis of Hi-C matrices at the interaction count level. The benchmark assesses the statistical methods, usability, and performance (in terms of precision and power) of these tools, using both real and simulated Hi-C data. Results reveal a striking variability in performance among the tools, highlighting the substantial impact of preprocessing filters and the difficulty all tools encounter in effectively controlling the false discovery rate across varying resolutions and chromosome sizes. Availability The complete benchmark is available at https://forgemia.inra.fr/scales/replication-chrocodiff using processed data deposited at https://doi.org/10.57745/LR0W9R. Contact nathalie.vialaneix@inrae.fr <<<

Weiran Feng, Erik Ladewig, Matthew Lange, Nazifa Salsabeel, Huiyong Zhao, Young Sun Lee, Anuradha Gopalan, Hanzhi Luo, Wenfei Kang, Ning Fan, Eric Rosiek, Elisa de Stanchina, Yu Chen, Brett S. Carver, Christina S. Leslie, Charles L. Sawyers <<<

Abstract Despite the high prevalence of ERG transcription factor translocations in prostate cancer, the mechanism of tumorigenicity remains poorly understood. Using lineage tracing, we find the tumor-initiating activity of ERG resides in a subpopulation of murine basal cells that coexpress luminal genes (BasalLum) and not in the larger population of ERG+ luminal cells. Upon ERG activation, BasalLum cells give rise to highly proliferative intermediate (IM) cells with stem-like features that coexpress basal, luminal, hillock and club marker genes, before transitioning to Krt8+ luminal cells. Transcriptomic analysis of ERG+ human prostate cancers confirms the presence of rare ERG+ BasalLum cells, as well as IM cells whose presence is associated with a worse prognosis. Single-cell analysis revealed a chromatin state in ERG+ IM cells enriched for STAT3 transcription factor binding sites and elevated expression of the KMT2A/MLL1 and DOT1L, all three of which are essential for ERG-driven tumorigenicity in vivo. In addition to providing translational opportunities, this work illustrates how single-cell approaches combined with lineage tracing can identify cancer vulnerabilities not evident from bulk analysis. <<<

Abstract Stress is widely considered to negatively impact hippocampal function, thus impairing episodic memory. However, the hippocampus is not merely the seat of episodic memory. Rather, it also (via distinct circuitry) supports statistical learning. On the basis of rodent work suggesting that stress may impair the hippocampal pathway involved in episodic memory while sparing or enhancing the pathway involved in statistical learning, we developed a behavioral experiment to investigate the effects of acute stress on both episodic memory and statistical learning in humans. Participants were randomly assigned to one of three conditions: stress (socially evaluated cold pressor) immediately before learning, stress ∼15 min before learning, or no stress. In the learning task, participants viewed a series of trial-unique scenes (allowing for episodic encoding of each image) in which certain scene categories reliably followed one another (allowing for statistical learning of associations between paired categories). Memory was assessed 24 hr later to isolate stress effects on encoding/learning rather than retrieval. We found modest support for our hypothesis that acute stress can amplify statistical learning: Only participants stressed ∼15 min in advance exhibited reliable evidence of learning across multiple measures. Furthermore, stress-induced cortisol levels predicted statistical learning retention 24 hr later. In contrast, episodic memory did not differ by stress condition, although we did find preliminary evidence that acute stress promoted memory for statistically predictable information and attenuated competition between statistical and episodic encoding. Together, these findings provide initial insights into how stress may differentially modulate learning processes within the hippocampus. <<<

Clinical trials show that strategies to elicit the production of broadly neutralizing antibodies have a promising start <<<