标题： 可解释的人工智能在基因组学中的应用：使用专家混合模型预测转录因子结合位点 显示英文标题

标题： Explainable AI in Genomics: Transcription Factor Binding Site Prediction with Mixture of Experts

摘要： 转录因子结合位点（TFBS）预测对于理解基因调控和各种生物过程至关重要。 本研究引入了一种新的专家混合（MoE）方法用于TFBS预测，该方法整合了多个预训练的卷积神经网络（CNN）模型，每个模型专门针对不同的TFBS模式。 我们在分布内和分布外（OOD）数据集上对我们的MoE模型与单独的专家模型进行了性能评估，使用六个随机选择的转录因子（TFs）进行OOD测试。 我们的结果表明，MoE模型在多种TF结合位点上表现出具有竞争力或更优的性能，特别是在分布外场景中表现尤为突出。 方差分析（ANOVA）统计检验确认了这些性能差异的显著性。 此外，我们引入了ShiftSmooth，这是一种新颖的归因映射技术，通过考虑输入序列的小幅移动，提供了更稳健的模型可解释性。 通过全面的可解释性分析，我们展示了 ShiftSmooth在基序发现和定位方面相比传统的Vanilla梯度方法具有更优的归因效果。 我们的工作为TFBS预测提供了一种高效、可推广且可解释的解决方案，有望在基因组生物学中带来新的发现，并推进我们对转录调控的理解。 显示英文摘要

摘要： Transcription Factor Binding Site (TFBS) prediction is crucial for understanding gene regulation and various biological processes. This study introduces a novel Mixture of Experts (MoE) approach for TFBS prediction, integrating multiple pre-trained Convolutional Neural Network (CNN) models, each specializing in different TFBS patterns. We evaluate the performance of our MoE model against individual expert models on both in-distribution and out-of-distribution (OOD) datasets, using six randomly selected transcription factors (TFs) for OOD testing. Our results demonstrate that the MoE model achieves competitive or superior performance across diverse TF binding sites, particularly excelling in OOD scenarios. The Analysis of Variance (ANOVA) statistical test confirms the significance of these performance differences. Additionally, we introduce ShiftSmooth, a novel attribution mapping technique that provides more robust model interpretability by considering small shifts in input sequences. Through comprehensive explainability analysis, we show that ShiftSmooth offers superior attribution for motif discovery and localization compared to traditional Vanilla Gradient methods. Our work presents an efficient, generalizable, and interpretable solution for TFBS prediction, potentially enabling new discoveries in genome biology and advancing our understanding of transcriptional regulation.