基于多源数据融合的复杂疾病建模方法研究

Research on Complex Disease Modeling Based on Multi-source Data Fusion

复杂疾病(如癌症)严重危害人类身体健康,其发生和发展往往是多因素共同作用的结果,如何清楚地认识疾病的内在致病机理是当前研究的重点内容.目前,随着高通量技术的发展,大量的生物数据涌现,因此,以数据为驱动,通过对生物系统的多源数据融合来实现复杂疾病的建模成为了一个新的研究热点.由于生物系统的多源数据往往在数据分布、尺度和格式等方面存在巨大差异,这使得如何有效地融合数据成为建模过程中的重点.本文从多源数据融合的角度,分别对疾病亚型识别、疾病相关模块挖掘和样本特异性疾病相关标志物识别3个方面对人类复杂疾病的建模方法进行讨论,力图为研究人员在人类复杂疾病的数据建模方面提供建议.

Complex diseases such as cancer,seriously endanger human health,their occurrence and development are often the result of multiple factors.How to clearly understand the internal pathogenesis of diseases is the focus of current research.Now,with the development of high throughput technology,a large number of biological data emerge.Therefore,it has become a new research hotspot to model complex diseases by multi-source data fusion of biological systems driven by data.However,because of the huge differences in data distribution,scale and format of multi-source data in biological system,how to effectively integrate data has become the focus of modeling process.From the perspective of multiple data fusion,we discuss the modeling methods of human complex diseases from three aspects:disease subtype recognition,disease-related module mining and sample specific disease-related marker recognition which try to provide suggestions for researchers in data modeling of complex human diseases.