<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[HalfPageTheorem的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/HalfPageTheorem</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; HalfPageTheorem]]></copyright><item><title><![CDATA[环肽设计AI方法综述]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/160243168</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/160243168</guid><author>HalfPageTheorem</author><pubDate>Fri, 17 Apr 2026 11:01:16 +0800</pubDate><description><![CDATA[环肽（cyclic peptides）作为一类兼具小分子药物稳定性与大分子生物制剂特异性的治疗模态，近年来在药物开发领域获得广泛关注。其中，头尾相连环肽，因其封闭骨架赋予的结构刚性与抗蛋白酶水解能力，展现出显著优于线性肽的药代动力学特性。]]></description><category></category></item><item><title><![CDATA[RFantibody：原子级精准抗体设计，现已上线 SciMiner！]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/157397433</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/157397433</guid><author>HalfPageTheorem</author><pubDate>Mon, 26 Jan 2026 16:54:41 +0800</pubDate><description><![CDATA[RFantibody 实现表位定向、原子级精准的抗体从头设计；SciMiner 现已上线，支持快速生成高质量候选。]]></description><category></category></item><item><title><![CDATA[AI赋能下的蛋白设计：RFdiffusion vs. BindCraft，谁是下一代研发利器？]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/157396786</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/157396786</guid><author>HalfPageTheorem</author><pubDate>Mon, 26 Jan 2026 16:37:45 +0800</pubDate><description><![CDATA[SciMiner 平台上对比 RFDiffusion 与 BindCraft 在 PD-L1 结合体设计中的表现与优劣。]]></description><category></category></item><item><title><![CDATA[告别“刻舟求剑”，PackDock如何搞定蛋白质的“七十二变”？]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/157252588</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/157252588</guid><author>HalfPageTheorem</author><pubDate>Thu, 22 Jan 2026 14:14:15 +0800</pubDate><description><![CDATA[PackDock 结合生成式 AI 与物理先验攻克蛋白受体柔性对接难题，现已入驻 SciMiner 平台，提供即开即用的云端侧链建模与构象分析体验。]]></description><category></category></item><item><title><![CDATA[18%精度跃迁，7倍效率加速！PBCNet2.0 开启 AI 驱动结合亲和力预测的新纪元]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/157249134</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/157249134</guid><author>HalfPageTheorem</author><pubDate>Thu, 22 Jan 2026 11:04:55 +0800</pubDate><description><![CDATA[PBCNet2.0 已正式上线 SciMiner 平台。该算法基于等变神经网络与860万海量数据，攻克了亲和力预测中精度与效率的矛盾。其预测精度提升 18%，研发提效达 718%，且具备卓越的耐药性突变预测能力，为药物研发提供既快又准的“AI原生驱动”方案。]]></description><category></category></item><item><title><![CDATA[Graph-pKa ：基于多实例学习与图神经网络的 pKₐ 预测新方法]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156768288</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156768288</guid><author>HalfPageTheorem</author><pubDate>Fri, 09 Jan 2026 17:08:36 +0800</pubDate><description><![CDATA[Graph-pKa 升级上线！基于 GNN 与多实例学习，实现高精度宏观预测与原子级微观 pKₐ 解析。]]></description><category></category></item><item><title><![CDATA[【重磅】蛋白设计进入“全原子生成”时代，BoltzGen 正式登陆 SciMiner 平台！]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156713495</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156713495</guid><author>HalfPageTheorem</author><pubDate>Thu, 08 Jan 2026 10:11:51 +0800</pubDate><description><![CDATA[BoltzGen 全原子扩散统一设计与折叠，SciMiner 平台化部署，提供多入口 Binder 生成体验。]]></description><category></category></item><item><title><![CDATA[【科研加速】告别环境配置地狱：SciMiner 深度集成 ProteinMPNN 全家桶，蛋白设计从此一键开启！]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156686983</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156686983</guid><author>HalfPageTheorem</author><pubDate>Thu, 08 Jan 2026 09:43:06 +0800</pubDate><description><![CDATA[SciMiner 深度集成 ProteinMPNN 全家桶，通过可视化交互与全链条工作流，破解环境配置难题，实现蛋白质序列的一键式高效设计与精准筛选。]]></description><category></category></item><item><title><![CDATA[当 AI 学会“造酶” ：RFdiffusion 的三次飞跃 ]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156647380</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156647380</guid><author>HalfPageTheorem</author><pubDate>Wed, 07 Jan 2026 13:56:41 +0800</pubDate><description><![CDATA[RFdiffusion 引领蛋白设计革命，SciMiner 让最前沿的“AI 造蛋白”真正走进科研日常。]]></description><category></category></item><item><title><![CDATA[Pocket2Mol 与 MolCRAFT：生成式分子设计的新路径]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156686528</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156686528</guid><author>HalfPageTheorem</author><pubDate>Wed, 07 Jan 2026 13:50:13 +0800</pubDate><description><![CDATA[Pocket2Mol 与 MolCRAFT 为分子设计提供新路径：前者逐步生成匹配口袋的小分子，后者在连续空间非自回归建模，实现更优结合与结构稳定性。]]></description><category></category></item><item><title><![CDATA[重磅！RFdiffusion3.0 正式上线，这波性能飞升你绝对不能错过！]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156645157</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156645157</guid><author>HalfPageTheorem</author><pubDate>Tue, 06 Jan 2026 13:38:33 +0800</pubDate><description><![CDATA[RFdiffusion3(RFD3)实现重大突破，将蛋白设计从骨架层面提升至原子级精度。该模型通过多分子共扩散技术，能同步优化蛋白质与配体/DNA的空间结构，支持蛋白-蛋白、蛋白-DNA、蛋白-小分子等多种相互作用设计。相比前代，RFD3采用全新Transformer-U-Net混合架构，计算效率提升10倍，并在实验验证中展现出优异的功能性。Sciminer平台已集成RFD3全流程设计工具，为科研人员提供从生成到验证的一站式解决方案，标志着蛋白设计进入"理解化学"的新阶段。]]></description><category></category></item><item><title><![CDATA[NA-MPNN 登陆 SciMiner：RNA 结构设计与 protein–DNA 特异性预测开启 3D 条件化新范式]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156644958</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/156644958</guid><author>HalfPageTheorem</author><pubDate>Tue, 06 Jan 2026 13:17:59 +0800</pubDate><description><![CDATA[NA-MPNN 正式登陆 SciMiner 平台，通过 3D 结构条件化生成技术，攻克核酸逆折叠难题，实现 RNA/DNA 序列的高效精准设计与蛋白复合物优化。]]></description><category></category></item><item><title><![CDATA[AI 赋能蛋白质工程：VenusFactory 开启“智能设计”新时代]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/154493362</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/154493362</guid><author>HalfPageTheorem</author><pubDate>Thu, 06 Nov 2025 16:54:41 +0800</pubDate><description><![CDATA[蛋白质工程（Protein Engineering）是生命科学与生物技术的核心支柱，广泛应用于抗体药物、疫苗、工业酶、合成生物学元件等领域。蛋白质的设计与优化能力，直接决定了研发效率与成果质量。然而，蛋白质的。以 300 个氨基酸组成的蛋白质为例，其理论组合数超过 20^300，远超任何实验室的筛选能力。传统方法依赖实验筛选与结构解析，往往需耗费数月乃至数年才能获得优化版本。以抗体药物为例，从发现到上市平均需 10 年，投入超过 10 亿美元，其中筛选与优化阶段往往占总成本的 60% +。]]></description><category></category></item><item><title><![CDATA[AI不止会预测结构，它还能“想”出分子]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/154491670</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/154491670</guid><author>HalfPageTheorem</author><pubDate>Thu, 06 Nov 2025 16:02:45 +0800</pubDate><description><![CDATA[Assay2Mol 会用大语言模型（ChatGPT 或 DeepSeek）理解语义，自动检索并学习与该靶点相关的活性分子，再生成新的候选结构。输入蛋白结构和一个参考小分子，模型会用小分子帮助定位口袋位置与化学环境，并在此基础上生成新的结构变体，快速探索结合位点周围的设计空间。👉 适合：当你只有靶点信息，还没有结构数据时，Assay2Mol 能帮你从“文字”直接到“分子”。👉 适合：当你只有靶点信息，还没有结构数据时，Assay2Mol 能帮你从“文字”直接到“分子”。从文本到结构，从定位到结合，]]></description><category></category></item><item><title><![CDATA[科研人都懂的“小麻烦”，SciMiner解决了。]]></title><link>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/154402645</link><guid>https://blog.youkuaiyun.com/HalfPageTheorem/article/details/154402645</guid><author>HalfPageTheorem</author><pubDate>Tue, 04 Nov 2025 17:26:36 +0800</pubDate><description><![CDATA[科研文件格式转换神器，一键搞定]]></description><category></category></item></channel></rss>