Jiayang Shi, Cuo Mei, Fengyong Ge, Qingliang Hu, Xinwei Ban, Ran Xia, Peiyong Xin, Shujing Cheng, Gaohua Zhang, Jiawei Nie, Shiqi Zhang, Xiaowei Ma, Yi Wang, Jinfang Chu, Yuhang Chen, Bing Wang, Weihua Wu, Jiayang Li, Qi Xie, Feifei Yu <<<

Large Language Models (LLMs) and LLM-based agents show great promise inaccelerating scientific research. Existing benchmarks for measuring thispotential and guiding future development continue to evolve from pure recalland rote knowledge tasks, towards more practical work such as literature reviewand experimental planning. Bioinformatics is a domain where fully autonomousAI-driven discovery may be near, but no extensive benchmarks for measuringprogress have been introduced to date. We therefore present the BioinformaticsBenchmark (BixBench), a dataset comprising over 50 real-world scenarios ofpractical biological data analysis with nearly 300 associated open-answerquestions designed to measure the ability of LLM-based agents to explorebiological datasets, perform long, multi-step analytical trajectories, andinterpret the nuanced results of those analyses. We evaluate the performance oftwo frontier LLMs (GPT-4o and Claude 3.5 Sonnet) using a custom agent frameworkwe open source. We find that even the latest frontier models only achieve 17%accuracy in the open-answer regime, and no better than random in amultiple-choice setting. By exposing the current limitations of frontiermodels, we hope BixBench can spur the development of agents capable ofconducting rigorous bioinformatic analysis and accelerate scientific discovery. <<<

Image inversion is a fundamental task in generative models, aiming to mapimages back to their latent representations to enable downstream applicationssuch as editing, restoration, and style transfer. This paper provides acomprehensive review of the latest advancements in image inversion techniques,focusing on two main paradigms: Generative Adversarial Network (GAN) inversionand diffusion model inversion. We categorize these techniques based on theiroptimization methods. For GAN inversion, we systematically classify existingmethods into encoder-based approaches, latent optimization approaches, andhybrid approaches, analyzing their theoretical foundations, technicalinnovations, and practical trade-offs. For diffusion model inversion, weexplore training-free strategies, fine-tuning methods, and the design ofadditional trainable modules, highlighting their unique advantages andlimitations. Additionally, we discuss several popular downstream applicationsand emerging applications beyond image tasks, identifying current challengesand future research directions. By synthesizing the latest developments, thispaper aims to provide researchers and practitioners with a valuable referenceresource, promoting further advancements in the field of image inversion. Wekeep track of the latest works at https://github.com/RyanChenYN/ImageInversion <<<

Clare Puttick, Thomas P. Jones, Michelle M. Leung, Felipe Galvez-Cancino, Jiali Liu, Manuel Varas-Godoy, Andrew Rowan, Oriol Pich, Carlos Martinez-Ruiz, Robert Bentham, Krijn K. Dijkstra, James R. M. Black, Rachel Rosenthal, Nnennaya Kanu, Kevin Litchfield, Roberto Salgado, David A. Moore, Peter Van Loo, Mariam Jamal-Hanjani, Sergio A. Quezada, , Heather Cheyne, Mohammed Khalil, Shirley Richardson, Tracey Cruickshank, Eric Lim, Hugo J. W. L. Aerts, Tom L. Kaufmann, Matthew R. Huska, Babu Naidu, Gareth A. Wilson, Rachel Rosenthal, Andrew Rowan, Chris Bailey, Claudia Lee, Emma Colliver, Katey S. S. Enfield, Mark S. Hill, Mihaela Angelova, Oriol Pich, Dhruva Biswas, Clare Puttick, Roberto Vendramin, Cian Murphy, Maria Zagorulya, Thomas P. Jones, Michelle M. Leung, Nicholas McGranahan, Carla Castignani, Elizabeth Larose Cadieux, Jeanette Kittel, Kerstin Haase, Kexin Koh, Rachel Scott, Gurdeep Matharu, Jacqui A. Shaw, Allan Hackshaw, Camilla Pilotti, Rachel Leslie, Anne-Marie Hacker, Sean Smith, Aoife Walker, Christopher Abbosh, Corentin Richard, Cristina Naceur-Lombardelli, Francisco Gimeno-Valiente, Krupa Thakkar, Mariana Werner Sunderland, Monica Sivakumar, Nnennaya Kanu, Ieva Usaite, Sadegh Saghafinia, Selvaraju Veeriah, Sharon Vanloo, Bushra Mussa, Michalina Magala, Elizabeth Keene, Emilia L. Lim, James R. sM Black, Maise Al Bakir, Ariana Huebner, Kristiana Grigoriadis, Takahiro Karasaki, Alexander M. Frankell, Crispin T. Hiley, Sophia Ward, Sian Harries, Olivia Lucas, David A. Moore, Nicolai J. Birkbak, Carlos Martínez-Ruiz, Kerstin Thol, Robert Bentham, Wing Kin Liu, Abigail Bunkum, Sonya Hessey, Martin D. Forster, Siow Ming Lee, Mariam Jamal-Hanjani, Despoina Karagianni, Sergio A. Quezada, Supreet Kaur Bola, Kevin Litchfield, Charles Swanton, John Le Quesne, Khalid AbdulJabbar, Catarina Veiga, Simone Zaccaria, Jonathan Tugwood, Caroline Dive, Zoltan Szallasi, Miklos Diossy, Teresa Marafioti, Elaine Borg, Mary Falzon, Reena Khiroya, Peter Van Loo, Karl S. Peggs, Gillian Price, Gary Royle, Charles-Antoine Collins-Fekete, Dionysis Papadatos-Pastos, James Wilson, Tanya Ahmad, Sarah Benafif, Judith Cave, Keith M. Kerr, Thomas B. K. Watkins, Roberto Salgado, Alexander James Procter, Asia Ahmed, Magali N. Taylor, Arjun Nair, David Lawrence, Davide Patrini, Colin R. Lindsay, Fiona H. Blackhall, Yvonne Summers, Matthew G. Krebs, Emma Nye, Richard Kevin Stone, Hanyun Zhang, Jerome Nicod, Alan Kirk, Mo Asif, Rocco Bilancia, Nikos Kostoulas, Jennifer Whiteley, Mathew Thomas, Akshay J. Patel, David Chuter, Mairead MacKenzie, Roland F. Schwarz, Andrew Kidd, Francesco Fraioli, Paul Ashford, Zoltan Kaplar, Jonas Demeulemeester, Claire Wilson, Michael J. Shackcloth, Sam M. Janes, Neal Navani, Ricky M. Thakrar, Angela Leek, Jack Davies Hodgkinson, Nicola Totton, Antonio Paiva-Correia, Stephan Beck, Miljana Tanic, Craig Dick, Lily Robinson, Peter Russell, Paulo De Sousa, Simon Jordan, Alexandra Rice, Hilgardt Raubenheimer, Harshil Bhayani, Lyn Ambrose, Anand Devaraj, Hemangi Chavan, Sofina Begum, Silviu I. Buderi, Daniel Kaniu, Mpho Malima, Sarah Booth, Nadia Fernandes, Pratibha Shah, Chiara Proli, Andrew G. Nicholson, Ekaterini Boleti, Madeleine Hewish, Kevin G. Blyth, Jason F. Lester, Anshuman Chaturvedi, Pedro Oliveira, Katherine D. Brown, Mathew Carter, Alastair Magness, Clare E. Weeden, Eva Grönroos, Jacki Goldman, Mickael Escudero, Philip Hobson, Stefan Boeing, Tamara Denner, Vittorio Barbè, Wei-Ting Lu, William Hill, Yutaka Naito, Zoe Ramsden, George Kassiotis, Imran Noorani, Anca Grapa, Aiman Alzetani, Yinyin Yuan, Xiaoxi Pan, Jack French, Kayleigh Gilbert, Angela Dwornik, Angeliki Karamani, Benny Chain, David R. Pearce, Felip Gálvez-Cancino, Georgia Stavrou, Gerasimos-Theodoros Mastrokalos, Helen L. Lowe, Ignacio Garcia Matos, James L. Reading, John A. Hartley, Kayalvizhi Selvaraju, Kezhong Chen, Leah Ensell, Mansi Shah, Maria Litovchenko, Piotr Pawlik, Samuel Gamble, Seng Kuong Anakin Ung, Victoria Spanswick, Yin Wu, Jayant K. Rane, Othman Al-Sawaf, Olga Chervova, Emilie Martinoni Hoogenboom, Fleur Monk, James W. Holding, Junaid Choudhary, Kunal Bhakhri, Pat Gorman, Robert C. M. Stephens, Maria Chiara Pisciella, Steve Bandula, Yien Ning Sophia Wong, Aya Osman, Mandeesh Sangha, Gerald Langman, Helen Shackleford, Madava Djearaman, Gary Middleton, Serena Chee, Patricia Georg, Amrita Bajaj, Apostolos Nakas, Azmina Sodha-Ramdeen, Mohamad Tufail, Molly Scotland, Rebecca Boyles, Sridhar Rathinam, Domenic Marrone, Sean Dulloo, Dean A. Fennell, Sarah Danson, Elaine Smith, Eustace Fontaine, Felice Granato, Juliette Novasio, Kendadai Rammohan, Leena Joseph, Paul Bishop, Rajesh Shah, Vijay Joshi, Philip Crosbie, Charles Swanton, Nicholas McGranahan <<<

AbstractDisruption of the class I human leukocyte antigen (HLA) molecules has important implications for immune evasion and tumor evolution. We developed major histocompatibility complex loss of heterozygosity (LOH), allele-specific mutation and measurement of expression and repression (MHC Hammer). We identified extensive variability in HLA allelic expression and pervasive HLA alternative splicing in normal lung and breast tissue. In lung TRACERx and lung and breast TCGA cohorts, 61% of lung adenocarcinoma (LUAD), 76% of lung squamous cell carcinoma (LUSC) and 35% of estrogen receptor-positive (ER+) cancers harbored class I HLA transcriptional repression, while HLA tumor-enriched alternative splicing occurred in 31%, 11% and 15% of LUAD, LUSC and ER+ cancers. Consistent with the importance of HLA dysfunction in tumor evolution, in LUADs, HLA LOH was associated with metastasis and LUAD primary tumor regions seeding a metastasis had a lower effective neoantigen burden than non-seeding regions. These data highlight the extent and importance of HLA transcriptomic disruption, including repression and alternative splicing in cancer evolution. <<<

Abstract Accurate prediction of enzyme kinetic parameters is crucial for enzyme exploration and modification. Existing models face the problem of either low accuracy or poor generalization ability due to overfitting. In this work, we first developed unbiased datasets to evaluate the actual performance of these methods and proposed a deep learning model, CataPro, based on pre-trained models and molecular fingerprints to predict turnover number (k c a t ), Michaelis constant (K m ), and catalytic efficiency (k c a t /K m ). Compared with previous baseline models, CataPro demonstrates clearly enhanced accuracy and generalization ability on the unbiased datasets. In a representational enzyme mining project, by combining CataPro with traditional methods, we identified an enzyme (SsCSO) with 19.53 times increased activity compared to the initial enzyme (CSO2) and then successfully engineered it to improve its activity by 3.34 times. This reveals the high potential of CataPro as an effective tool for future enzyme discovery and modification. <<<

AbstractMulti‐omics usually refers to the crossover application of multiple high‐throughput screening technologies represented by genomics, transcriptomics, single‐cell transcriptomics, proteomics and metabolomics, spatial transcriptomics, and so on, which play a great role in promoting the study of human diseases. Most of the current reviews focus on describing the development of multi‐omics technologies, data integration, and application to a particular disease; however, few of them provide a comprehensive and systematic introduction of multi‐omics. This review outlines the existing technical categories of multi‐omics, cautions for experimental design, focuses on the integrated analysis methods of multi‐omics, especially the approach of machine learning and deep learning in multi‐omics data integration and the corresponding tools, and the application of multi‐omics in medical researches (e.g., cancer, neurodegenerative diseases, aging, and drug target discovery) as well as the corresponding open‐source analysis tools and databases, and finally, discusses the challenges and future directions of multi‐omics integration and application in precision medicine. With the development of high‐throughput technologies and data integration algorithms, as important directions of multi‐omics for future disease research, single‐cell multi‐omics and spatial multi‐omics also provided a detailed introduction. This review will provide important guidance for researchers, especially who are just entering into multi‐omics medical research. <<<

ABSTRACTHow the prefrontal cortex contributes to working memory remains controversial, as theories differ in their emphasis on its role in storing memories versus controlling their content. To adjudicate between these competing ideas, we tested how perturbations to the human (both sexes) lateral prefrontal cortex impact the storage and control aspects of working memory during a task that requires human subjects to allocate resources to memory items based on their behavioral priority. Our computational model made a strong prediction that disruption of this control process would counterintuitively improve memory for low-priority items. Remarkably, transcranial magnetic stimulation of retinotopically-defined superior precentral sulcus, but not intraparietal sulcus, unbalanced the prioritization of resources, improving memory for low-priority items as predicted by the model. Therefore, these results provide direct causal support for models in which the prefrontal cortex controls the allocation of resources that support working memory, rather than simply storing the features of memoranda.SIGNIFICANCE STATEMENTAlthough higher-order cognition depends on working memory, the resources that support our memory are severely limited in capacity. To mitigate this limitation, we allocate memory resources according to the behavioral relevance of items. Nonetheless, the neural basis of these abilities remain unclear. Here, we tested the hypothesis that a region in lateral prefrontal cortex controls prioritization in working memory. Indeed, perturbing this region with transcranial magnetic stimulation disrupted the prioritization of working memory resources. Our results provide causal evidence for the hypothesis that prefrontal cortex primarily controls the allocation of memory resources, rather than storing the contents of working memory. <<<

Xunliang Jiang, Jun Liu, Ke Wang, Jianyong Sun, Huilong Yin, Yu Jiang, Yongkang Liu, Ningbo Wang, Xiaochen Ding, Pu Gao, Lin Li, Xiang Zhang, Jipeng Li, Rui Zhang <<<

The realm of board games presents a challenging domain for the application of artificial intelligence (AI), given their vast state-action space and inherent complexity. This paper explores the development of a proficient AI for Connect Four using DeepMind's AlphaZero algorithm. The algorithm employs a policy-value network for concurrent prediction of action probabilities and state values, and Monte Carlo Tree Search (MCTS) for decision-making, guided by the policy-value network. Through extensive self-play and data augmentation, our AI learns without the need for explicit prior knowledge. Our experiment demonstrated that the AI player showed significant capability in playing Connect Four, exhibiting strategic decision-making that sometimes-surpassed human performance. These results underline the potential of deep reinforcement learning in advancing AI performance in complex board games. <<<

BACKGROUND: Artesunate (ART) has been used extensively as anti-malarial drugs worldwide. Besides, it has also been reported to have anti-cancer activities. This study was aimed to explore the anti-cancer activity of ART in combination with cisplatin (CIS) on A549 cells.METHODS: Cells were cultured with different concentrations of ART and/or CIS for 24, 48, or 72 h to test the anti-proliferative effects by CCK-8 assay. Colony formation assay and EdU staining were also performed. TUNEL staining was used to illustrate the morphologic changes. Cell cycle and apoptosis were determined by flow cytometry assay, and Western blot analysis was conducted to detect the expression of apoptosis- and proliferation-related proteins. Caspase activities were determined by colorimetric assay kit. Moreover, the synergistic effect of ART with CIS in A549 cell xenograft model was also determined.RESULTS: ART significantly inhibited cell proliferation in dose- and time-dependent manners. Collectively, the combination treatment remarkably decreased colony formation rates and increased the rates of TUNEL-positive cells compared with mono-treatment. Mechanistically, the combination treatment influenced the expression of Bcl-2, Bax, p-P53, Caspase-3/7, Caspase-9, CyclinB1, P34, P21, and synergistically regulated the activity of P38/JNK/ERK1/2 MAPK pathway. In mice A549 xenograft tumors, the combination strategy significantly increased the anti-cancer efficacy of ART and CIS alone, consistent with the in vitro observations.CONCLUSIONS: ART exhibited significant anti-tumor effect on A549 cells and this efficiency could be enhanced by combination with CIS. <<<

Mirazul Islam, Yilin Yang, Alan J. Simmons, Vishal M. Shah, Krushna Pavan Musale, Yanwen Xu, Naila Tasneem, Zhengyi Chen, Linh T. Trinh, Paola Molina, Marisol A. Ramirez-Solano, Iannish D. Sadien, Jinzhuang Dou, Andrea Rolong, Ken Chen, Mark A. Magnuson, Jeffrey C. Rathmell, Ian G. Macara, Douglas J. Winton, Qi Liu, Hamim Zafar, Reza Kalhor, George M. Church, Martha J. Shrubsole, Robert J. Coffey, Ken S. Lau <<<

AbstractAntigen discovery technologies have largely focused on major histocompatibility complex (MHC) class I-restricted human T cell receptors (TCRs), leaving methods for MHC class II-restricted and mouse TCR reactivities relatively undeveloped. Here we present TCR mapping of antigenic peptides (TCR-MAP), an antigen discovery method that uses a synthetic TCR-stimulated circuit in immortalized T cells to activate sortase-mediated tagging of engineered antigen-presenting cells (APCs) expressing processed peptides on MHCs. Live, tagged APCs can be directly purified for deconvolution by sequencing, enabling TCRs with unknown specificity to be queried against barcoded peptide libraries in a pooled screening context. TCR-MAP accurately captures self-reactivities or viral reactivities with high throughput and sensitivity for both MHC class I-restricted and class II-restricted TCRs. We elucidate problematic cross-reactivities of clinical TCRs targeting the cancer/testis melanoma-associated antigen A3 and discover targets of myocarditis-inciting autoreactive T cells in mice. TCR-MAP has the potential to accelerate T cell antigen discovery efforts in the context of cancer, infectious disease and autoimmunity. <<<

Large Vision-Language-Action (VLA) models, leveraging powerful pre trainedVision-Language Models (VLMs) backends, have shown promise in robotic controldue to their impressive generalization ability. However, the success comes at acost. Their reliance on VLM backends with billions of parameters leads to highcomputational costs and inference latency, limiting the testing scenarios tomainly quasi-static tasks and hindering performance in dynamic tasks requiringrapid interactions. To address these limitations, this paper proposes HiRT, aHierarchical Robot Transformer framework that enables flexible frequency andperformance trade-off. HiRT keeps VLMs running at low frequencies to capturetemporarily invariant features while enabling real-time interaction through ahigh-frequency vision-based policy guided by the slowly updated features.Experiment results in both simulation and real-world settings demonstratesignificant improvements over baseline methods. Empirically, in static tasks,we double the control frequency and achieve comparable success rates.Additionally, on novel real-world dynamic ma nipulation tasks which arechallenging for previous VLA models, HiRT improves the success rate from 48% to75%. <<<

Practical relevance: The ‘2022 AAFP/ISFM Cat Friendly Veterinary Interaction Guidelines: Approach and Handling Techniques’ (hereafter the ‘Cat Friendly Veterinary Interaction Guidelines’) support veterinary professionals with feline interactions and handling to reduce the impact of fear and other protective (negative) emotions, in so doing enhancing feline welfare and In implementing these Guidelines, team satisfaction and cat caregiver confidence in the veterinary team will increase as the result of efficient examinations, better experience, more reliable diagnostic testing and improved feline wellbeing. Veterinary professionals will learn the importance of understanding and appropriately responding to the current emotional state of the cat and tailoring each visit to the individual. Clinical challenges: Cats have evolved with emotions and behaviors that are necessary for their survival as both a predator and prey species. A clinical setting and the required examinations and procedures to meet their physical health needs can result in behavioral responses to protective emotions. Cat friendly interactions require understanding, interpreting and appropriately responding to cats’ emotional states and giving them a perceived sense of control while performing the required assessment. Evidence base: These Guidelines have been created by a Task Force of experts convened by the American Association of Feline Practitioners and the International Society of Feline Medicine, based on an extensive literature review and, where evidence is lacking, the authors’ experience. Endorsements: These Guidelines have been endorsed by a number of groups and organizations, as detailed on page 1127 and at catvets.com/interactions and icatcare.org/cat-friendly-guidelines . <<<

To train generalist robot policies, machine learning methods often require asubstantial amount of expert human teleoperation data. An ideal robot forhumans collecting data is one that closely mimics them: bimanual arms anddexterous hands. However, creating such a bimanual teleoperation system withover 50 DoF is a significant challenge. To address this, we introduce Bidex, anextremely dexterous, low-cost, low-latency and portable bimanual dexterousteleoperation system which relies on motion capture gloves and teacher arms. Wecompare Bidex to a Vision Pro teleoperation system and a SteamVR system andfind Bidex to produce better quality data for more complex tasks at a fasterrate. Additionally, we show Bidex operating a mobile bimanual robot for in thewild tasks. The robot hands (5k USD) and teleoperation system (7k USD) isreadily reproducible and can be used on many robot arms including two xArms(16k USD). Website at https://bidex-teleop.github.io/ <<<

Isabel Deisenhofer, Jean-Paul Albenque, Sonia Busch, Edouard Gitenay, Stavros E. Mountantonakis, Antoine Roux, Jerome Horvilleur, Babe Bakouboula, Saumil Oza, Selim Abbey, Guillaume Theodore, Antoine Lepillier, Yves Guyomar, Francis Bessiere, Jaap Jan Smit, Theophile Mohr Durdez, Paola Milpied, Anthony Appetiti, Daniel Guerrero, Tom De Potter, Christian De Chillou, Seth Goldbarg, Atul Verma, John D. Hummel, , Jaap Jan Smit, Tom De Potter, Christian De Chillou <<<

Abstract Although pulmonary vein isolation (PVI) has become the cornerstone ablation procedure for atrial fibrillation (AF), the optimal ablation procedure for persistent and long-standing persistent AF remains elusive. Targeting spatio-temporal electrogram dispersion in a tailored procedure has been suggested as a potentially beneficial alternative to a conventional PVI-only procedure. In this multicenter, randomized, controlled, double-blind, superiority trial, patients with drug-refractory persistent AF were randomly assigned to a tailored ablation procedure targeting areas of spatio-temporal dispersion, as detected by an artificial intelligence (AI) algorithm, in addition to PVI (tailored arm, n = 187, 23% women) or to a conventional PVI-only procedure (anatomical arm, n = 183, 19% women). The primary efficacy endpoint was freedom from documented AF with or without antiarrhythmic drugs at 12 months after a single ablation procedure. Secondary endpoints included freedom from any atrial arrhythmic events, and the secondary composite safety endpoint consisted of death, cerebrovascular events, or treatment-related serious adverse events. One year post-procedure, the trial met its primary efficacy endpoint, which was achieved in 88% of patients in the tailored arm compared with 70% of patients in the anatomical arm (log-rank P < 0.0001 for superiority). However, no significant difference between arms was observed for the freedom from any atrial arrhythmia endpoint after one ablation. The safety endpoint did not differ between arms, with procedure and ablation times being twice as long in the tailored arm. These results show that AI-guided ablation of spatio-temporal dispersion areas in addition to PVI is superior to PVI alone in eliminating AF at 1-year follow-up in patients with persistent and long-standing persistent AF. Ablation of subsequent organized atrial tachycardias may be needed to maintain sinus rhythm long term. ClinicalTrials.gov identifier: NCT04702451. <<<

Zhiyi Chen, Yancheng Tang, Xuerong Liu, Wei Li, Yuanyuan Hu, Bowen Hu, Ting Xu, Rong Zhang, Lei Xia, Jing-Xuan Zhang, Zhibing Xiao, Ji Chen, Zhengzhi Feng, Yuan Zhou, Qinghua He, Jiang Qiu, Xu Lei, Hong Chen, Shaozheng Qin, Tingyong Feng <<<