检索与竞争：蛋白质基础模型如何启动蛋白质
Retrieval and competition: how a protein foundation model starts a protein
AI总结：
通过追踪ESM2-8M预测蛋白质起始甲硫氨酸的计算路径，发现模型依赖位置先验检索而非直接识别，揭示了模型置信度与生物学证据之间的脱节。
摘要：
蛋白质语言模型越来越多地用于指导实验和临床决策，但通常不清楚一个自信的预测是反映了对生物学证据的识别还是对统计默认值的检索。我们针对一个近乎普遍的生物学规则——蛋白质以甲硫氨酸起始——通过追踪ESM2-8M产生该预测的计算路径来检验这一区别。模型并未检测到掩码位置的甲硫氨酸。相反，它通过跨层组装的特定位置查询，从序列起始标记处的参考表示中检索出有利于甲硫氨酸的信号，最终输出通过与上下文相关电路的竞争而出现。为了理解位置信息如何到达读出端，我们引入了旋转频率带内注意力分数的范数-方向分解。位置编码通过分布在各个频带中的查询范数和角度对齐的耦合变化来运作。对于真实N端不是甲硫氨酸的序列（此时生物学问题至关重要），模型仍然预测甲硫氨酸。这不是由意外机制产生的正确预测，而是匹配统计平均值的位置先验检索电路的输出，在生物学偏离平均值的地方失败。区分这两者需要在单个电路、频率带和查询组成的层面上进行解析，这表明在生物学风险更高的预测中，机制验证将是必要且具有挑战性的。即使对于最简单的生物学规则，模型的预测也是通过分布式计算电路而非直接识别来介导的，这表明任务复杂性的增加将进一步模糊模型置信度与潜在生物学证据之间的关系。
英文摘要：
Protein language models are increasingly used to guide experimental and clinical decisions, yet it is often unclear whether a confident prediction reflects recognition of biological evidence or retrieval of a statistical default. We examine this distinction for a near-universal biological rule, that proteins begin with methionine, by tracing the computational pathway through which ESM2-8M produces this prediction. The model does not detect methionine at the masked position. Instead, it retrieves a methionine-favouring signal from a reference representation at the beginning-of-sequence token via a position-specific query assembled across layers, with the final output emerging through competition with context-dependent circuits. To understand how positional information reaches the readout, we introduce a norm-direction decomposition of attention scores within rotary frequency bands. Positional encoding operates through coupled changes in query norm and angular alignment distributed across these bands. On sequences whose true N-terminus is not methionine, where the biological question matters, the model predicts methionine anyway. This is not a correct prediction produced by an unexpected mechanism, but the output of a positional-prior retrieval circuit that matches the statistical average and fails where biology diverges from it. Distinguishing the two requires resolution at the level of individual circuits, frequency bands, and query composition, suggesting that mechanistic verification will be necessary, and challenging, for predictions where the biological stakes are higher. Even for the simplest biological rule, the model's prediction is mediated by a distributed computational circuit rather than direct recognition, suggesting that increasing task complexity will further obscure the relationship between model confidence and underlying biological evidence.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
