当编辑流即编辑跳跃：复现 Edit Flows 与 EvoFlows
When Edit Flows are Edit Jumps: replicating Edit Flows and EvoFlows
- Visium
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本文复现Edit Flows与EvoFlows，提出首个开源实现EditJumps，用1.66M同源对训练通用抗体编辑器，实现零样本编辑，并揭示开放代码对生成生物学的重要性。
AI中文摘要：
抗体先导物优化要求对现有候选物进行少量、有界的编辑：替换，也包括插入和删除。基于编辑的生成模型是唯一能够在无需预先固定编辑位置、编辑次数或输出长度的情况下分配此类编辑预算的模型。然而，现有方法 Edit Flows 和 EvoFlows 未发布代码或完整的训练规格。在此，我们表明这两种方法遵循相同的基本过程——编辑在连续时间内以学习到的速率逐个触发——即有限序列上生成器匹配的纯跳跃情形。通过 EditJumps，我们首次提供了该框架的开源实现，使用在 1.66M 个 Observed Antibody Space 同源对上训练的单一通用抗体编辑器，为种子序列提出类同源变体，可零样本编辑未见过的先导物，无需原始方法所需的逐家族重训练。从头复现该系统揭示了开放代码对生成生物学至关重要的原因：调和已发表的编辑分布需要对一个未记录的速率缩放超参数进行逆向工程，该超参数决定了实际实现的突变计数。此外，我们表明已发表的评估指标对参考样本量高度敏感，经常导致方法排名翻转。我们在以下网址发布我们的完整代码库、自动化测试套件和配置：this https URL
英文摘要：
Antibody lead optimization calls for a small, bounded set of edits to an existing candidate: substitutions, but also insertions and deletions. Edit-based generative models are the only ones that allocate such an edit budget without fixing the edit positions, the edit count, or the output length in advance. However, the existing approaches Edit Flows and EvoFlows did not release code or complete training specifications. Here, we show that both methods follow the same underlying process -- edits firing one at a time, at learned rates, in continuous time -- the pure-jump case of generator matching over finite sequences. With EditJumps we introduce the first open implementation of this framework, with a single generalist antibody editor trained on 1.66M Observed Antibody Space homolog pairs to propose homolog-like variants of a seed sequence, editing unseen leads zero-shot, without the per-family retraining original approaches require. Replicating this system from scratch exposes why open code is essential for generative biology: reconciling published edit distributions required reverse-engineering an undocumented rate-scaling hyperparameter that dictates realized mutation counts. Moreover, we show that published evaluation metrics are highly sensitive to reference sample size, frequently flipping method rankings. We release our full codebase, automated test suite, and configurations at: https://github.com/VisiumCH/editjumps
