因果电路追踪揭示单细胞基础模型中的不同计算架构：抑制主导、生物一致性与跨模型收敛
Causal Circuit Tracing Reveals Distinct Computational Architectures in Single-Cell Foundation Models: Inhibitory Dominance, Biological Coherence, and Cross-Model Convergence
AI总结：
该研究通过因果电路追踪揭示单细胞基础模型中抑制主导、生物一致性和跨模型收敛的计算架构特征。
摘要：
动机：稀疏自编码器（SAEs）将基础模型激活分解为可解释的特征，但生物基础模型中网络深度的因果特征到特征交互仍未知。结果：我们通过消去SAE特征并测量下游响应引入因果电路追踪，并将其应用于Geneformer V2-316M和scGPT全人类模型，在四个条件下（96,892条边，80,191次前向传递）。两种模型显示约53%的生物一致性和65至89%的抑制主导，无论架构和细胞类型如何。scGPT产生更强的效果（均绝对d=1.40 vs. 1.05），动态更平衡。跨模型共识产生1,142对保守的领域对（10.6倍富集，p < 0.001）。与疾病相关的领域更可能成为共识。基因层面的CRISPRi验证显示56.4%的定向准确性，证实共表达而非因果编码。
英文摘要：
Motivation: Sparse autoencoders (SAEs) decompose foundation model activations into interpretable features, but causal feature-to-feature interactions across network depth remain unknown for biological foundation models. Results: We introduce causal circuit tracing by ablating SAE features and measuring downstream responses, and apply it to Geneformer V2-316M and scGPT whole-human across four conditions (96,892 edges, 80,191 forward passes). Both models show approximately 53 percent biological coherence and 65 to 89 percent inhibitory dominance, invariant to architecture and cell type. scGPT produces stronger effects (mean absolute d = 1.40 vs. 1.05) with more balanced dynamics. Cross-model consensus yields 1,142 conserved domain pairs (10.6x enrichment, p < 0.001). Disease-associated domains are 3.59x more likely to be consensus. Gene-level CRISPRi validation shows 56.4 percent directional accuracy, confirming co-expression rather than causal encoding.
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。
