AI+Bio 24/12/31 文献速递|Alignab结合预训练语言模型、扩散模型及帕累托能量对齐方法,有效设计自然抗体
作者：微信文章关注AI+Biology线报，每日推送AI+Biology当日更新的最新文章和资讯，获取领域最新的技术进展
Deepmind推出Alphafold3官方教程，从原理到实操带你精通结构预测重磅｜Deepmind推出Alphafold3官方教程，从原理到实操带你精通结构预测数据决定成败：机器学习在小分子药物研发的未来｜Nature Computational Science最新综述
阿斯利康重磅｜抗体设计中的生成模型全面评测
一击即中！BindCraft 实现蛋白binder的one-shot设计（附完整protocol
1. Alignab: Pareto-Optimal Energy Alignment for Designing Nature-Like Antibodies
期刊: arXiv/cs.lg
链接: https://arxiv.org/abs/2412.20984
总结:
该研究提出了一种三阶段的抗体序列-结构联合设计框架，结合语言模型预训练、扩散模型优化以及帕累托能量对齐方法，有效设计了自然类抗体，提高了结合位点的吸引力和功能性。
摘要:
We present a three-stage framework for training deep learning models specializing in antibody sequence-structure co-design. We first pre-train a language model using millions of antibody sequence data. Then, we employ the learned representations to guide the training of a diffusion model for joint optimization over both sequence and structure of antibodies.
2. ProtCLIP: Function-Informed Protein Multi-Modal Learning
期刊: arXiv
链接: https://arxiv.org/abs/2412.20014
总结:
ProtCLIP提出了一种基于蛋白功能的预训练范式，结合大规模蛋白-文本数据集ProtAnno和细粒度功能建模策略，在多模态蛋白表征任务中取得了显著性能提升，是蛋白多模态基础模型的重要进展。
摘要:
Multi-modality pre-training paradigm that aligns protein sequences and biological descriptions has learned general protein representations and achieved promising performance in various downstream applications. However, these works were still unable to replicate the extraordinary success of language-supervised visual foundation models.
3. Peptune: De Novo Generation of Therapeutic Peptides with Multi-Objective-Guided Discrete Diffusion
期刊: arXiv/q-bio.bm cs.ai
链接: https://arxiv.org/abs/2412.17780
总结:
Peptune通过蒙特卡洛树搜索（MCTS）引导的离散扩散模型，成功优化生成满足多种治疗属性的多样化肽类分子，为多目标优化任务提供了强大的设计工具。
摘要:
Peptide therapeutics, a major class of medicines, have achieved remarkable success across diseases such as diabetes and cancer. Despite their success, designing peptides that satisfy multiple conflicting objectives remains a major challenge.
4. Navigating Chemical-Linguistic Sharing Space with Heterogeneous Molecular Encoding
期刊: arXiv/cs.ce
链接: https://arxiv.org/abs/2412.20888
总结:
该研究提出HME框架，结合异构分子编码和多任务数据集，显著提升化学-语言共享空间的探索能力，为化学领域基础研究与应用提供了新思路。
摘要:
Chemical language models (CLMs) are prominent for their effectiveness in exploring chemical space and enabling molecular engineering. However, while exploring chemical-linguistic space, CLMs suffer from the gap between natural language and molecular representations.
5. Protscan: Modeling and Prediction of RNA-Protein Interactions
期刊: arXiv/q-bio.bm
链接: https://arxiv.org/abs/2412.20933
总结:
Protscan基于一致性核化SGD回归模型，提供了一种预测RNA-蛋白相互作用的新工具，克服了实验噪声的影响，能够广泛应用于全转录组尺度的研究中。
摘要:
CLIP-seq methods are valuable techniques to experimentally determine transcriptome-wide binding sites of RNA-binding proteins. Despite the constant improvement of such techniques, the results are affected by various types of noise.
6. Parallel DNA Sequence Alignment on High-Performance Systems with CUDA and MPI
期刊: arXiv/cs.dc
链接: https://arxiv.org/abs/2412.21103
总结:
结合CUDA和MPI的混合实现方式，显著加速了DNA序列比对流程，为大规模生物信息数据集的处理提供了高效解决方案。
摘要:
Sequence alignment is a cornerstone of bioinformatics, widely used to identify similarities between DNA, RNA, and protein sequences. The Needleman-Wunsch algorithm remains a robust and accurate method for global sequence alignment.
7. SFE-Net: Harnessing Biological Principles of Differential Gene Expression for Improved Feature Selection in Deep Learning Networks
期刊: arXiv/cs.mm
链接: https://arxiv.org/abs/2412.20799
总结:
SFE-Net受生物学差异基因表达启发，提出了一种动态特征选择机制，有效提高了模型在深度伪造检测任务中的泛化能力。
摘要:
In the realm of deepfake detection, the challenge of adapting to various synthesis methodologies such as FaceSwap, DeepFakes, Face2Face, and NeuralTextures significantly impacts the performance of traditional machine learning models.
8. Predicting Non-Coding RNA Function Using Artificial Intelligence
期刊: bioRxiv
链接: https://www.biorxiv.org/content/10.1101/2024.12.30.630736
总结:
该研究利用自然语言处理方法构建了非编码RNA-表型关系数据集和关联模型，为非编码RNA功能研究提供了全面的分析工具。
摘要:
Non-coding RNAs (ncRNAs) represent the majority of human gene products, and are involved in various important biological processes. It is then of pivotal importance to aggregate and summarize the existing information.
9. TCPGdb: A Comprehensive T Cell Perturbation Genomics Database for Identification of Critical T Cell Regulators
期刊: bioRxiv
链接: https://www.biorxiv.org/content/10.1101/2024.12.30.630773
总结:
TCPGdb汇总多项T细胞基因干扰研究数据，提供系统性分析工具，有助于发现T细胞治疗潜在靶点，提高CAR-T细胞疗效。
摘要:
CAR-T therapies utilizing T cells engineered with chimeric antigen receptors (CARs) have revolutionized the treatment of hematologic and immune-related malignancies.
10. Graph Network-Based Analysis of Disease-Gene-Drug Associations
期刊: bioRxiv
链接: https://www.biorxiv.org/content/10.1101/2024.12.30.630746
总结:
ZGNT基于图神经网络实现无监督疾病-药物关联预测，为疾病治疗策略的开发提供了新的视角和工具。
摘要:
ZGNT, an innovative, novel workflow for zero-shot learning in drug repurposing that leverages meta-path graph neural network transformers. This method infers disease-drug relationships indirectly.
DiffSBDD 是一种基于对称性扩散模型的新方法，通过 3D 条件生成问题，拓展了结构药物设计的适用性，为药物生成提供了更广泛的解决方案。
页:
[1]