We construct $\varepsilon$-approximate unitary $k$-designs on $n$ qubits incircuit depth $O(\log k \log \log n k / \varepsilon)$. The depth isexponentially improved over all known results in all three parameters $n$, $k$,$\varepsilon$. We further show that each dependence is optimal up toexponentially smaller factors. Our construction uses $\tilde{{O}}(nk)$ ancillaqubits and ${O}(nk)$ bits of randomness, which are also optimal up to $\log(nk)$ factors. An alternative construction achieves a smaller ancilla count$\tilde{{O}}(n)$ with circuit depth ${O}(k \log \log nk/\varepsilon)$. Toachieve these efficient unitary designs, we introduce a highly-structuredrandom unitary ensemble that leverages long-range two-qubit gates and low-depthimplementations of random classical hash functions. We also develop a newanalytical framework for bounding errors in quantum experiments involving manyqueries to random unitaries. As an illustration of this framework'sversatility, we provide a succinct alternative proof of the existence ofpseudorandom unitaries. <<<

AbstractBackgroundAdvances in cancer biology are increasingly dependent on integration of heterogeneous datasets. Large-scale efforts have systematically mapped many aspects of cancer cell biology; however, it remains challenging for individual scientists to effectively integrate and understand this data.ResultsWe have developed a new data retrieval and indexing framework that allows us to integrate publicly available data from different sources and to combine publicly available data with new or bespoke datasets. Our approach, which we have named the cancer data integrator (CanDI), is straightforward to implement, is well documented, and is continuously updated which should enable individual users to take full advantage of efforts to map cancer cell biology. We show that CanDI empowered testable hypotheses of new synthetic lethal gene pairs, genes associated with sex disparity, and immunotherapy targets in cancer.ConclusionsCanDI provides a flexible approach for large-scale data integration in cancer research enabling rapid generation of hypotheses. The CanDI data integrator is available athttps://github.com/GilbertLabUCSF/CanDI. <<<

A major challenge in the burgeoning field of quantum simulation forhigh-energy physics is the realization of scalable $2+1$D lattice gaugetheories on state-of-the-art quantum hardware, which is an essential steptowards the overarching goal of probing $3+1$D quantum chromodynamics on aquantum computer. Despite great progress, current experimental implementationsof $2+1$D lattice gauge theories are mostly restricted to relatively smallsystem sizes and two-level representations of the gauge and electric fields.Here, we propose a resource-efficient method for quantum simulating $2+1$Dspin-$S$ $\mathrm{U}(1)$ quantum link lattice gauge theories with dynamicalmatter using qudit-based quantum processors. By integrating out the matterfields through Gauss's law, we reformulate the quantum link model in a purelyspin picture compatible with qudit encoding across arbitrary spatialdimensions, eliminating the need for ancillary qubits and reducing resourceoverhead. Focusing first on the spin-$1/2$ case, we construct explicit circuitsfor the full Hamiltonian and demonstrate through numerical simulations that thefirst-order Trotterized circuits accurately capture the quench dynamics even inthe presence of realistic noise levels. Additionally, we introduce a generalmethod for constructing coupling-term circuits for higher-spin representations$S>1/2$. Compared to conventional qubit encodings, our framework significantlyreduces the number of quantum resources and gate count. Our approachsignificantly enhances scalability and fidelity for probing nonequilibriumphenomena in higher-dimensional lattice gauge theories, and is readily amenableto implementation on state-of-the-art qudit platforms. <<<

BACKGROUND Massively parallel DNA sequencing of cell-free fetal DNA from maternal blood can detect fetal chromosomal abnormalities. Although existing algorithms focus on the detection of fetal trisomy 21 (T21), these same algorithms have difficulty detecting trisomy 18 (T18). METHODS Blood samples were collected from 1014 patients at 13 US clinic locations before they underwent an invasive prenatal procedure. All samples were processed to plasma, and the DNA extracted from 119 samples underwent massively parallel DNA sequencing. Fifty-three sequenced samples came from women with an abnormal fetal karyotype. To minimize the intra- and interrun sequencing variation, we developed an optimized algorithm by using normalized chromosome values (NCVs) from the sequencing data on a training set of 71 samples with 26 abnormal karyotypes. The classification process was then evaluated on an independent test set of 48 samples with 27 abnormal karyotypes. RESULTS Mapped sites for chromosomes of interest in the sequencing data from the training set were normalized individually by calculating the ratio of the number of sites on the specified chromosome to the number of sites observed on an optimized normalizing chromosome (or chromosome set). Threshold values for trisomy or sex chromosome classification were then established for all chromosomes of interest, and a classification schema was defined. Sequencing of the independent test set led to 100% correct classification of T21 (13 of 13) and T18 (8 of 8) samples. Other chromosomal abnormalities were also identified. CONCLUSION Massively parallel sequencing is capable of detecting multiple fetal chromosomal abnormalities from maternal plasma when an optimized algorithm is used. <<<

Design of T cell receptors by artificial intelligence is poised to accelerate cancer immunotherapy <<<

Matteo Ridelfi, Giacomo Vezzani, Emanuele Roscioli, Giampiero Batani, Elena Boero, Francesco Nannini, Eleonora Marini, Ushasi Bhaumik, Girmay Desalegn, Ornella Serpino, Antonio Molinaro, Ida Paciello, Chiara Mugnaini, Concetta De Santi, Giuseppe Maccari, Antonella De Rosa, Silvia Valensin, Assia Duatti, Roberta Di Benedetto, Maria Michelina Raso, Renzo Alfini, Chiara Sammicheli, Simona Tavarini, Christina Quigley, Audino Podda, Anna Kabanova, Emanuele Andreano, Dario Cardamone, Gianmarco Gasperini, Francesca Necchi, Mariagrazia Pizza, Laura B. Martin, Robert W. Frenck, Alexandre Grassart, Alba Silipo, Marcela F. Pasetti, Francesco Berlanda Scorza, Carlo Giannelli, Omar Rossi, Francesca Micoli, Claudia Sala, Rino Rappuoli <<<

Shigellosis is a global public health challenge that mostly affects low- and middle-income countries and causes considerable morbidity and mortality among children under 5 y of age. Multi- and extensively drug-resistant Shigella sonnei strains associated with recent outbreaks in high-income countries exacerbate the problem and have prompted the World Health Organization to include Shigella spp. among the high-risk pathogens for which novel prophylactic and therapeutic tools are urgently needed. Among the most promising and cutting-edge solutions, monoclonal antibodies are gaining considerable attention in the infectious diseases field. Here, we report the discovery of human monoclonal antibodies against S. sonnei , a species whose prevalence is constantly increasing worldwide and is associated with frequent drug-resistant infections. We isolated antibodies generated in response to an experimental S. sonnei vaccine followed by a controlled human infection and screened them by using a panel of high-throughput assays. We identified a molecule which exhibited potent bactericidal activity in vitro, inhibition of invasion of epithelial cells and conferred full protection from S. sonnei infection in vivo. Overall, our study provides a candidate antibody that can rapidly progress to industrial development for application as a prophylactic, therapeutic, and diagnostic tool against shigellosis. <<<

Marco Todesco, Gregory L. Owens, Natalia Bercovich, Jean-Sébastien Légaré, Shaghayegh Soudi, Dylan O. Burge, Kaichi Huang, Katherine L. Ostevik, Emily B. M. Drummond, Ivana Imerovski, Kathryn Lande, Mariana A. Pascual-Robles, Mihir Nanavati, Mojtaba Jahani, Winnie Cheung, S. Evan Staton, Stéphane Muños, Rasmus Nielsen, Lisa A. Donovan, John M. Burke, Sam Yeaman, Loren H. Rieseberg <<<

Sarah Lewis, Tim Hempel, José Jiménez-Luna, Michael Gastegger, Yu Xie, Andrew Y. K. Foong, Victor García Satorras, Osama Abdin, Bastiaan S. Veeling, Iryna Zaporozhets, Yaoyi Chen, Soojung Yang, Adam E. Foster, Arne Schneuing, Jigyasa Nigam, Federico Barbero, Vincent Stimper, Andrew Campbell, Jason Yim, Marten Lienen, Yu Shi, Shuxin Zheng, Hannes Schulz, Usman Munir, Roberto Sordillo, Ryota Tomioka, Cecilia Clementi, Frank Noé <<<

Following the sequence and structure revolutions, predicting functionally relevant protein structure changes at scale remains an outstanding challenge. We introduce BioEmu, a deep learning system that emulates protein equilibrium ensembles by generating thousands of statistically independent structures per hour on a single GPU. BioEmu integrates over 200 milliseconds of molecular dynamics (MD) simulations, static structures and experimental protein stabilities using novel training algorithms. It captures diverse functional motions—including cryptic pocket formation, local unfolding, and domain rearrangements—and predicts relative free energies with 1 kcal/mol accuracy compared to millisecond-scale MD and experimental data. BioEmu provides mechanistic insights by jointly modelling structural ensembles and thermodynamic properties. This approach amortizes the cost of MD and experimental data generation, demonstrating a scalable path toward understanding and designing protein function. <<<

Petri Pölönen, Danika Di Giacomo, Anna Eames Seffernick, Abdelrahman Elsayed, Shunsuke Kimura, Francesca Benini, Lindsey E. Montefiori, Brent L. Wood, Jason Xu, Changya Chen, Zhongshan Cheng, Haley Newman, Jason Myers, Ilaria Iacobucci, Elizabeth Li, Jonathan Sussman, Dale Hedges, Yawei Hui, Caroline Diorio, Lahari Uppuluri, David Frank, Yiping Fan, Yunchao Chang, Soheil Meshinchi, Rhonda Ries, Rawan Shraim, Alexander Li, Kathrin M. Bernt, Meenakshi Devidas, Stuart S. Winter, Kimberly P. Dunsmore, Hiroto Inaba, William L. Carroll, Nilsa C. Ramirez, Aaron H. Phillips, Richard W. Kriwacki, Jun J. Yang, Tiffaney L. Vincent, Yaqi Zhao, Pankaj S. Ghate, Jian Wang, Colleen Reilly, Xin Zhou, Mathijs A. Sanders, Junko Takita, Motohiro Kato, Nao Takasugi, Bill H. Chang, Richard D. Press, Mignon Loh, Evadnie Rampersaud, Elizabeth Raetz, Stephen P. Hunger, Kai Tan, Ti-Cheng Chang, Gang Wu, Stanley B. Pounds, Charles G. Mullighan, David T. Teachey <<<

Siegfried Schloissnig, Samarendra Pani, Jana Ebler, Carsten Hain, Vasiliki Tsapalou, Arda Söylev, Patrick Hüther, Hufsah Ashraf, Timofey Prodanov, Mila Asparuhova, Hugo Magalhães, Wolfram Höps, Jesus Emiliano Sotelo-Fonseca, Tomas Fitzgerald, Walter Santana-Garcia, Ricardo Moreira-Pinhal, Sarah Hunt, Francy J. Pérez-Llanos, Tassilo Erik Wollenweber, Sugirthan Sivalingam, Dagmar Wieczorek, Mario Cáceres, Christian Gilissen, Ewan Birney, Zhihao Ding, Jan Nygaard Jensen, Nikhil Podduturi, Jan Stutzki, Bernardo Rodriguez-Martin, Tobias Rausch, Tobias Marschall, Jan O. Korbel <<<

Abstract Genomic structural variants (SVs) contribute substantially to genetic diversity and human diseases1–4, yet remain under-characterized in population-scale cohorts5. Here we conducted long-read sequencing6 in 1,019 humans to construct an intermediate-coverage resource covering 26 populations from the 1000 Genomes Project. Integrating linear and graph genome-based analyses, we uncover over 100,000 sequence-resolved biallelic SVs and we genotype 300,000 multiallelic variable number of tandem repeats7, advancing SV characterization over short-read-based population-scale surveys3,4. We characterize deletions, duplications, insertions and inversions in distinct populations. Long interspersed nuclear element-1 (L1) and SINE-VNTR-Alu (SVA) retrotransposition activities mediate the transduction8,9 of unique sequence stretches in 5′ or 3′, depending on source mobile element class and locus. SV breakpoint analyses point to a spectrum of homology-mediated processes contributing to SV formation and recurrent deletion events. Our open-access resource underscores the value of long-read sequencing in advancing SV characterization and enables guiding variant prioritization in patient genomes. <<<

Games have played a pivotal role in advancing artificial intelligence, withAI agents using sophisticated techniques to compete. Despite the success ofneural network based game AIs, their performance often requires significantcomputational resources. In this paper, we present Rapfi, an efficient Gomokuagent that outperforms CNN-based agents in limited computation environments.Rapfi leverages a compact neural network with a pattern-based codebookdistilled from CNNs, and an incremental update scheme that minimizescomputation when input changes are minor. This new network uses computationthat is orders of magnitude less to reach a similar accuracy of much largerneural networks such as Resnet. Thanks to our incremental update scheme,depth-first search methods such as the alpha-beta search can be significantlyaccelerated. With a carefully tuned evaluation and search, Rapfi reachedstrength surpassing Katagomo, the strongest open-source Gomoku AI based onAlphaZero's algorithm, under limited computational resources where acceleratorslike GPUs are absent. Rapfi ranked first among 520 Gomoku agents on Botzone andwon the championship in GomoCup 2024. <<<

Sebastian Soyk, Zachary H. Lemmon, Matan Oved, Josef Fisher, Katie L. Liberatore, Soon Ju Park, Anna Goren, Ke Jiang, Alexis Ramos, Esther van der Knaap, Joyce Van Eck, Dani Zamir, Yuval Eshed, Zachary B. Lippman <<<

Ann C Huang, Tsung-Han S Hsieh, Jiang Zhu, Jackson Michuda, Ashton Teng, Soohong Kim, Elizabeth M Rumsey, Sharon K Lam, Ikenna Anigbogu, Philip Wright, Mohamed Ameen, Kwontae You, Christopher J Graves, Hyunsung John Kim, Adam J Litterman, Rene V Sit, Alex Blocker, Ci Chu <<<

AbstractThe rapid expansion of massively parallel sequencing technologies has enabled the development of foundation models to uncover novel biological findings. While these have the potential to significantly accelerate scientific discoveries by creating AI-driven virtual cell models, their progress has been greatly limited by the lack of large-scale high-quality perturbation data, which remains constrained due to scalability bottlenecks and assay variability. Here, we introduce “Fix-Cryopreserve-ScRNAseq” (FiCS) Perturb-seq, an industrialized platform for scalable Perturb-seq data generation. We demonstrate that FiCS Perturb-seq exhibits high sensitivity and low batch effects, effectively capturing perturbation-induced transcriptomic changes and recapitulating known biological pathways and protein complexes. In addition, we release X-Atlas: Orion edition (X-Atlas/Orion), the largest publicly available Perturb-seq atlas. This atlas, generated from two genome-wide FiCS Perturb-seq experiments targeting all human protein-coding genes, comprises eight million cells deeply sequenced to over 16,000 unique molecular identifiers (UMIs) per cell. Furthermore, we show that single guide RNA (sgRNA) abundance can serve as a proxy for gene knockdown (KD) efficacy. Leveraging the deep sequencing and substantial cell numbers per perturbation, we also show that stratification by sgRNA expression can reveal dose-dependent genetic effects. Taken together, we demonstrate that FiCS Perturb-seq is an efficient and scalable platform for high-throughput Perturb-seq screens. Through the release of X-Atlas/Orion, we highlight the potential of FiCS Perturb-seq to address current scalability and variability challenges in data generation, advance foundation model development that incorporates gene-dosage effects, and accelerate biological discoveries. <<<

Abstract Background and Hypothesis Stress sensitivity may represent an important target for resilience-promoting, preventive interventions. Resilience Training (RT) is a 4-session, group-based behavioral intervention, focusing on mindfulness-based skills, that leads to reductions in psychopathology. To investigate the mechanisms of RT, the current study tested whether acquiring the skills taught in RT leads to decreases in psychopathology via reductions in one manifestation of stress sensitivity, emotion reactivity, and associated changes in hippocampal-frontal connectivity. Study Design An open trial of RT was conducted in 103 non-help-seeking young adults with mild-to-moderate psychotic experiences (PEs) and/or symptoms of depression. Transdiagnostic symptoms, emotion reactivity, and mindfulness-related skills were measured, and, in a subset of participants (n = 41), resting-state, functional magnetic resonance imaging (fMRI) data were collected, before and after completion of RT. Study Results Replicating and extending findings of prior studies of RT, significant increases in mindfulness-related RT skills and significant decreases in transdiagnostic symptoms and emotion reactivity, as well as changes in hippocampal-frontal functional connectivity, were observed following RT (all P < .02). Mediation analyses revealed that associations between the acquisition of the RT skills and decreases in symptoms (P < .006) were fully mediated by the decrease in emotion reactivity, which was also correlated with a significant pre-to-post increase in hippocampal-dorsolateral prefrontal cortex connectivity (P = .032). Conclusions RT may lead to improvements in mental health by increasing the capacity to manage day-to-day stress. Future randomized controlled trials with extended follow-up can determine whether such improvements decrease the likelihood of developing disabling mental illnesses in transdiagnostically at-risk individuals. <<<