We make use of ideas from the theory of complex networks to implement a machine learning classification of human DNA methylation data, that carry signatures of cancer development. The data were obtained from patients with various kinds of cancers and represented as parenclictic networks, wherein nodes correspond to genes, and edges are weighted according to pairwise variation from control group subjects. We demonstrate that for the $10$ types of cancer under study, it is possible to obtain a high performance of binary classification between cancer-positive and negative samples based on network measures. Remarkably, an accuracy as high as $93-99\%$ is achieved with only $12$ network topology indices, in a dramatic reduction of complexity from the original $15295$ gene methylation levels. Moreover, it was found that the parenclictic networks are scale-free in cancer-negative subjects, and deviate from the power-law node degree distribution in cancer. The node centrality ranking and arising modular structure could provide insights into the systems biology of cancer.
Non-Invasive MGMT Status Prediction in GBM Cancer Using Magnetic Resonance Images (MRI) Radiomics Features: Univariate and Multivariate Machine Learning Radiogenomics Analysis
Kermanshah University of Medical Sciences(克尔曼沙阿医科大学);Rajaie Cardiovascular Medical and Research Center(拉贾伊心血管医疗与研究中心);Iran University of Medical Science(伊朗医科大学);The University of British Columbia(不列颠哥伦比亚大学);Kerman University(克尔曼大学);University of British Columbia(不列颠哥伦比亚大学)
背景和目的:本研究旨在通过MRI放射组学特征以及单因素和多因素分析来预测O-6甲基鸟嘌呤-DNA甲基转移酶(MGMT)基因启动子的甲基化状态。材料和方法:82例MGMT甲基化状态的患者纳入本研究。在MR图像的四个区域中手动分割肿瘤,a)整个肿瘤,b)活动/增强区域,c)坏死区域和d)水肿区域(E)。为每个患者提取了大约七千个放射组学特征。通过不同的机器学习算法,使用特征选择和分类器来预测MGMT状态。使用受试者工作特征(ROC)曲线的曲线下面积(AUC)进行模型评价。结果如下:关于单变量分析,发现使用4.5 mm Sigma拉普拉斯高斯滤波器(AUC:0.71(p值:0.002))的整个肿瘤段中的灰度共生矩阵(GLCM)的逆方差特征是最佳预测因子。对于多变量分析,在水肿区域中具有从模型选择特征选择器和筛选过滤器的决策树分类器具有最高性能(AUC:0.78),其次是在水肿区域中具有从模型选择特征选择器和筛选过滤器的Ada Boost分类器(AUC:0.74)。结论:这项研究表明,使用机器学习算法的放射组学是预测GBM癌症患者MGMT甲基化状态的可行的非侵入性方法关键词:放射组学,放射基因组学,GBM,MRI,MGMT
Identifying multi-omics interactions for lung cancer drug targets discovery using Kernel Machine Regression
Department of Information and Communication Technology, Mawlana Bhashani Science and Technology University, Santosh, Tangail, 1902, Dhaka, Bangladesh(信息与通信技术系,穆拉纳巴沙尼科学与技术大学);Department of Computer Science, Tulane University, New Orleans, LA, USA(计算机科学系,杜兰大学);Bioinformatics Institute (BII), Agency for Science, Technology and Research (A*STAR), 138632, Singapore(生物信息学研究所(BII),科技研究局(A*STAR));Tulane Center for Biomedical Informatics and Genomics, Deming Department of Medicine, Tulane University, New Orleans, LA 70112, USA(杜兰大学生物医学信息学与基因组学中心,医学部);Ochsner Center for Outcomes Research, Ochsner Research, New Orleans, LA 70121, USA(Ochsner研究中心,Ochsner研究)
Bayesian Genome- and Epigenome-wide Association Studies with Gene Level Dependence
高通量遗传和表观遗传数据经常被筛选与观察到的表型的关联。例如,人们可能希望测试数十万个遗传变异或DNA甲基化位点与疾病状态的关联。这些基因组变量可以自然地根据它们编码的基因以及其他标准进行分组。然而,在这种应用中的标准实践是独立筛选,具有对多重性的普遍校正。我们提出了一种贝叶斯方法,其中一个给定的基因组变量的关联的先验概率取决于它的基因,和基因特异性的概率建模非参数。这种分层模型允许适当的基因和全基因组多重性调整,并可以被纳入到各种贝叶斯关联筛选方法中,计算复杂性的增加可以忽略不计。我们描述了一个应用程序,以筛选低级别胶质瘤和多形性胶质母细胞瘤肿瘤样本之间的DNA甲基化差异的癌症基因组图谱。软件可通过https://github.com/lockEF/BayesianScreening上的BayesianScreening for R软件包获得。
Quantum AI for Cancer Diagnostic Biomarker Discovery
Department of Electrical and Computer Engineering, North Carolina State University, NC, USA(电气与计算机工程系,北卡罗来纳州立大学,NC,USA);Department of Biochemistry, Purdue University, IN, USA(生物化学系,普渡大学,IN,USA);Department of Chemistry, North Carolina State University, Raleigh, NC 27695(化学系,北卡罗来纳州立大学,雷德伍德,NC 27695)