[论文解读] Analysis of Extremely Obese Individuals Using Deep Learning Stacked Autoencoders and Genome-Wide Genetic Data

Casimiro Adays Curbelo Montañez, Paul Fergus+2 more|arXiv (Cornell University)|Apr 16, 2018

Genetic Associations and Epidemiology被引用 4

一句话总结

本研究提出一种基于堆叠自编码器（SAE）的深度学习方法，用于分析极重度肥胖个体的全基因组遗传数据，通过捕捉非线性SNP-SNP互作关系，克服了传统GWAS的局限性。该方法在分类极重度肥胖时，AUC达到0.975，基尼指数为0.950，显著优于线性GWAS模型。

ABSTRACT

The aetiology of polygenic obesity is multifactorial, which indicates that life-style and environmental factors may influence multiples genes to aggravate this disorder. Several low-risk single nucleotide polymorphisms (SNPs) have been associated with BMI. However, identified loci only explain a small proportion of the variation ob-served for this phenotype. The linear nature of genome wide association studies (GWAS) used to identify associations between genetic variants and the phenotype have had limited success in explaining the heritability variation of BMI and shown low predictive capacity in classification studies. GWAS ignores the epistatic interactions that less significant variants have on the phenotypic outcome. In this paper we utilise a novel deep learning-based methodology to reduce the high dimensional space in GWAS and find epistatic interactions between SNPs for classification purposes. SNPs were filtered based on the effects associations have with BMI. Since Bonferroni adjustment for multiple testing is highly conservative, an important proportion of SNPs involved in SNP-SNP interactions are ignored. Therefore, only SNPs with p-values < 1x10-2 were considered for subsequent epistasis analysis using stacked auto encoders (SAE). This allows the nonlinearity present in SNP-SNP interactions to be discovered through progressively smaller hidden layer units and to initialise a multi-layer feedforward artificial neural network (ANN) classifier. The classifier is fine-tuned to classify extremely obese and non-obese individuals. The best results were obtained with 2000 compressed units (SE=0.949153, SP=0.933014, Gini=0.949936, Lo-gloss=0.1956, AUC=0.97497 and MSE=0.054057). Using 50 compressed units it was possible to achieve (SE=0.785311, SP=0.799043, Gini=0.703566, Logloss=0.476864, AUC=0.85178 and MSE=0.156315).

研究动机与目标

• 为解决传统GWAS在解释BMI遗传力和分类极重度肥胖方面预测能力有限的问题。
• 识别线性GWAS因保守的多重检验校正而遗漏的SNP之间非线性上位互作关系。
• 利用堆叠自编码器对高维遗传数据进行降维，以提升分类性能。
• 构建一个深度学习流程，用于初始化并微调多层前馈神经网络以实现肥胖分类。
• 使用标准指标（如AUC、基尼指数和对数损失）在经过SNP p值<1×10⁻²筛选的数据集上评估模型性能。

提出的方法

• 基于p值<1×10⁻²筛选SNP，以降低维度，同时保留潜在相关变异。
• 应用堆叠自编码器（SAE）通过逐层编码和解码，学习SNP数据的分层非线性表示。
• 将SAE生成的压缩表示（2000和50个单元）作为多层前馈人工神经网络（ANN）分类器的输入。
• 通过反向传播对ANN分类器进行微调，以优化极重度肥胖与非肥胖个体的分类性能。
• 采用两阶段流程：首先通过SAE进行预训练以初始化权重，随后进行端到端微调以提升性能。
• 使用标准指标评估模型性能：AUC、基尼指数、敏感性（SE）、特异性（SP）、对数损失和均方误差（MSE）。

实验结果

研究问题

• RQ1 基于深度学习的降维方法（通过堆叠自编码器）是否能超越传统GWAS，在极重度肥胖分类方面实现性能提升？
• RQ2 非线性SNP-SNP互作在极重度肥胖个体BMI遗传力中贡献程度如何？
• RQ3 采用较不保守的p值阈值（p<1×10⁻²）过滤SNP，对上位互作检测有何影响？
• RQ4 在堆叠自编码器中，最优压缩表示大小（单元数量）是多少，可使极重度肥胖预测的分类准确率最大化？
• RQ5 与标准GWAS方法相比，SAE模型在AUC、基尼指数和对数损失等性能指标上的表现如何？

主要发现

• 2000个压缩单元的模型AUC达到0.97497，表明其在区分极重度肥胖与非肥胖个体方面具有优异的判别能力。
• 同一模型的基尼指数为0.949936，反映出其在预测排序方面表现强劲。
• 敏感性（SE）为0.949153，特异性（SP）为0.933014，表明真阳性率和真阴性率均较高。
• 对数损失最小化至0.1956，表明预测概率具有高度置信度和准确性。
• 即使仅使用50个压缩单元，模型仍实现AUC 0.85178和基尼指数0.703566，显示出在低维情况下的鲁棒性。
• 2000单元模型的均方误差（MSE）为0.054057，表明预测误差较低。