发表机构
Indian Institute of Technology Patna（印度理工学院巴特那分校）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
GenoMorph通过将基因组序列与潜在通路动态对齐，实现基于通路的疾病推理，在匿名基准上显著提升F1并降低延迟，优于依赖记忆关联的现有方法。
AI 中文摘要
大型语言模型（LLMs）在生物学推理方面展现了强大的能力；然而，基因组疾病推断在很大程度上仍依赖于记忆的基因-疾病关联，而非理解生物学通路。这种捷径学习削弱了鲁棒性和泛化能力，并且在分子标识符不可用时失效。我们提出了GenoMorph，一个多模态基因组推理框架，将疾病预测从关联性基因-疾病映射转向基于通路的推理。GenoMorph将冻结的DNA基础模型与问题条件的交叉注意力融合、自适应潜在推理（LatentSp）、用于迭代基因组证据再注入的残差推理门控，以及由分层最优传输（OT）正则化的拒绝采样微调相结合。GenoMorph并非学习直接的基因-疾病映射，而是将基因组序列表示与潜在通路动态对齐，使得推理轨迹在产生疾病预测之前遵循分子相互作用。LatentSp根据推理置信度动态分配计算，减少不必要的推理步骤并提高推理效率。我们进一步从京都基因与基因组百科全书（KEGG）构建了一个匿名基准，将每个基因和分子标识符替换为匿名符号，同时保留序列和通路拓扑，从而消除记忆捷径。GenoMorph将加权F1从0.7863（BioReason）提升至0.9412，而带有自适应潜在推理的拒绝采样微调将其推至0.9725，同时将延迟降低近60%。在匿名基准上，它达到了0.9465的F1，大幅优于先前系统，并确认了准确的疾病预测可以源于通路推理而非记忆的基因-疾病关联。
英文摘要
Large language models (LLMs) have demonstrated strong capabilities in biological reasoning; however, genomic disease inference remains largely dependent on memorized gene-disease associations rather than understanding biological pathways. This shortcut learning undermines robustness and generalization, and breaks down when molecular identifiers are unavailable. We present GenoMorph, a multimodal genomic reasoning framework that shifts disease prediction from associative gene-disease mapping toward pathway-grounded reasoning. GenoMorph couples a frozen DNA foundation model with question-conditioned cross-attention fusion, self-adaptive latent reasoning (LatentSp), a residual reasoning gate for iterative genomic evidence reinjection, and rejection sampling fine-tuning regularized by hierarchical optimal transport (OT). Rather than learning direct gene-disease mappings, GenoMorph aligns genomic sequence representations with latent pathway dynamics, enabling reasoning trajectories that follow molecular interactions before producing disease predictions. LatentSp dynamically allocates computation according to reasoning confidence, reducing unnecessary reasoning steps and improving inference efficiency. We further construct an anonymized benchmark from the Kyoto Encyclopedia of Genes and Genomes (KEGG), replacing every gene and molecular identifier with anonymous symbols while preserving sequences and pathway topology, thereby removing memorization shortcuts. GenoMorph raises the weighted F1 from 0.7863 (BioReason) to 0.9412, and rejection sampling fine-tuning with self-adaptive latent reasoning pushes it to 0.9725 while cutting latency nearly 60%. On the anonymized benchmark it reaches 0.9465 F1, substantially outperforming prior systems and confirming that accurate disease prediction can arise from pathway reasoning rather than memorized gene-disease associations.
