Detecting Alzheimer's Disease (AD) at the earliest possible stage is key in advancing AD prevention and treatment but is challenged by normal aging processes in addition to other confounding neurodegenerative diseases. Recent genome-wide association studies (GWAS) have identified associated alleles, but it has been difficult to transition from non-coding genetic variants to underlying mechanisms of AD. Here, we sought to reveal functional genetic variants and diagnostic biomarkers underlying AD using machine learning techniques. We first developed a Random Forest (RF) classifier using microarray gene expression data sampled from the peripheral blood of 744 participants in the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort. After initial feature selection, 5-fold cross-validation of the 100-gene RF classifier achieved an accuracy of 99.04%. The high accuracy of the RF classifier supports the possibility of a powerful and minimally invasive tool for screening of AD. Next, unsupervised clustering was used to validate and identify relationships among differentially expressed genes (DEGs) the RF selected revealing 3 distinct AD clusters. Results suggest downregulation of global sulfatase and oxidoreductase activities in AD through mutations in SUMF1 and SMOX respectively. Then, we used Greedy Fast Causal Inference (GFCI) to find potential causes of AD within DEGs. In the causal graph, HLA-DPB1 and CYP4A11 emerge as hub genes, furthering the discussion of the immune system's role in AD. Finally, we used Gene Set Enrichment Analysis (GSEA) to determine the biological pathways and processes underlying the DEGs that were highly correlated with AD. Cell activation in the immune system, glycosaminoglycan (GAG) binding, vascular dysfunction, oxidative stress, and the neuronal apoptotic process were revealed to be significantly enriched in AD. This study further advances the possibility of low-cost and noninvasive genetic screening for AD while also providing potential gene targets for further experimentation

中文翻译：

---

阿尔茨海默氏病的功能遗传生物标志物和外周血的基因表达

尽早发现阿尔茨海默氏病（AD）是推进AD预防和治疗的关键，但除其他混杂的神经退行性疾病外，正常的衰老过程也给它带来了挑战。最近的全基因组关联研究（GWAS）已经确定了相关的等位基因，但是很难从非编码遗传变异转变为AD的潜在机制。在这里，我们试图揭示使用机器学习技术的AD的功能遗传变异和诊断生物标记。我们首先使用微阵列基因表达数据开发了一个随机森林（RF）分类器，该数据是从阿尔茨海默氏病神经影像学倡议（ADNI）队列中744名参与者的外周血中取样的。选择初始功能后，100个基因的RF分类器的5倍交叉验证的准确性为99.04％。RF分类器的高精度支持使用功能强大且微创的工具来筛查AD的可能性。接下来，使用无监督聚类来验证和识别RF选择的差异表达基因（DEG）之间的关系，该基因揭示了3个不同的AD簇。结果表明，分别通过SUMF1和SMOX突变，AD中总体硫酸酯酶和氧化还原酶活性下调。然后，我们使用贪婪快速因果推断（GFCI）来发现DEG中AD的潜在原因。在因果图中，HLA-DPB1和CYP4A11作为枢纽基因出现，从而进一步讨论了免疫系统在AD中的作用。最后，我们使用基因集富集分析（GSEA）来确定与AD高度相关的DEG的生物学途径和过程。免疫系统中的细胞活化，糖胺聚糖（GAG）结合，血管功能障碍，氧化应激和神经元凋亡过程被发现在AD中显着丰富。这项研究进一步提高了低成本，无创性筛查AD的可能性，同时也为进一步的实验提供了潜在的基因靶标