基于稀疏表示的精神分裂症生物标记物筛选方法
发布时间：2019-01-01 19:56
【摘要】：精神分裂症是以思维、情感与行为的分裂为主要特征的一类复杂精神疾病,国内外大量研究发现遗传因素是该疾病发生的重要原因。为了能够从大量影像学和遗传学数据中找出精神分裂症相关的生物标记物,受稀疏表示的启发,提出一种基于稀疏表示的影像遗传学数据整合分析方法,并应用于对精神分裂症相关生物标记物的筛选。针对从208个样本中提取到的41 236组f MRI和722 177组SNP数据,通过对传统稀疏表示模型施加广义惩罚限制,然后对两类数据施加不同的权重因子α1、α2,并且使用不同的Lp(p=0、0.5、1)范数对模型分别求解,研究不同条件下两类数据的显著关联特征规律。结果发现,基因DAOA和HTR2A在下列多重情况下均被筛选出:一是f MRI数据的权重α1取0.35~0.8之间多个不同权重时,二是SNP数据的权重α2仅为0.2时,三是在L0、L0.5、L1等3种不同范数下。此外,在影像学数据方面,发现顶下缘角回脑区也与精神分裂症相关,此发现与先前精神分裂症的影像学研究结果一致。研究结果表明,将基于稀疏表示的影像遗传学数据整合分析方法应用于精神分裂症的生物标记物筛选是一个可行的方法,这为今后精神分裂症的影像遗传学研究提供了一种新的研究思路。
[Abstract]:Schizophrenia is a kind of complex mental disease characterized by the division of thinking, emotion and behavior. A large number of studies at home and abroad have found that genetic factors are the important cause of the disease. In order to find out the biomarkers related to schizophrenia from a large number of imaging and genetic data, and inspired by sparse representation, a method of image genetic data integration and analysis based on sparse representation is proposed. And applied to the screening of schizophrenia related biomarkers. For 41,236 sets of f MRI and 722,177 sets of SNP data extracted from 208 samples, a generalized penalty restriction is imposed on the traditional sparse representation model, and then different weighting factors 伪 1, 伪 2 are applied to the two types of data. Different Lp (p0. 0. 5) norm is used to solve the model respectively, and the significant correlation characteristics of the two kinds of data under different conditions are studied. The results showed that the gene DAOA and HTR2A were screened under the following multiple conditions: first, when the weight 伪 1 of f MRI data was between 0. 35 and 0. 8, the second was that the weight of SNP data was only 0. 2, and the third was in L0, L0. 5. L1 and other three different norms. In addition, in terms of imaging data, we found that the area of the inferior parietal angle gyrus is also associated with schizophrenia, which is consistent with the previous findings of the imaging study of schizophrenia. The results show that it is feasible to apply the image genetic data integration analysis method based on sparse representation to the screening of biomarkers in schizophrenia. This provides a new way to study the imaging genetics of schizophrenia in the future.
【作者单位】： 上海理工大学医疗器械与食品学院;
【基金】：国家自然科学基金(61101174) 上海理工大学微创基金项目(YS30809153)
【分类号】：R749.3
本文编号：2398048
[Abstract]:Schizophrenia is a kind of complex mental disease characterized by the division of thinking, emotion and behavior. A large number of studies at home and abroad have found that genetic factors are the important cause of the disease. In order to find out the biomarkers related to schizophrenia from a large number of imaging and genetic data, and inspired by sparse representation, a method of image genetic data integration and analysis based on sparse representation is proposed. And applied to the screening of schizophrenia related biomarkers. For 41,236 sets of f MRI and 722,177 sets of SNP data extracted from 208 samples, a generalized penalty restriction is imposed on the traditional sparse representation model, and then different weighting factors 伪 1, 伪 2 are applied to the two types of data. Different Lp (p0. 0. 5) norm is used to solve the model respectively, and the significant correlation characteristics of the two kinds of data under different conditions are studied. The results showed that the gene DAOA and HTR2A were screened under the following multiple conditions: first, when the weight 伪 1 of f MRI data was between 0. 35 and 0. 8, the second was that the weight of SNP data was only 0. 2, and the third was in L0, L0. 5. L1 and other three different norms. In addition, in terms of imaging data, we found that the area of the inferior parietal angle gyrus is also associated with schizophrenia, which is consistent with the previous findings of the imaging study of schizophrenia. The results show that it is feasible to apply the image genetic data integration analysis method based on sparse representation to the screening of biomarkers in schizophrenia. This provides a new way to study the imaging genetics of schizophrenia in the future.
【作者单位】： 上海理工大学医疗器械与食品学院;
【基金】：国家自然科学基金(61101174) 上海理工大学微创基金项目(YS30809153)
【分类号】：R749.3
【相似文献】
相关博士学位论文 前4条
1 陈新贵;精神分裂症幻听及其重复经颅磁刺激治疗的神经机制研究[D];安徽医科大学;2017年
2 杨楹;儿童精神分裂症5年预后及社会功能康复影响因素的前瞻性研究[D];山东大学;2017年
3 朱佳佳;基于MRI新技术的精神分裂症及其核心症状的神经机制研究[D];天津医科大学;2017年
4 付程;精神分裂症相关蛋白dysbindin在细胞核内降解以及对ERK1/2功能调控的研究[D];中国科学技术大学;2014年
相关硕士学位论文 前1条
1 姚星星;精神分裂症患者用药模式及医疗资源和费用分析[D];天津大学;2016年
，本文编号：2398048
本文链接：https://www.wllwen.com/yixuelunwen/jsb/2398048.html
最近更新
教材专著
