面向复杂疾病的序列标记物发现及其动态迁移研究

The search and discovery of the biomarkers of complex diseases at the molecular level can help study the mechanisms of the occurrence and process of complex diseases. The biomarkers play a import role in early diagnosis, stage determining, drug target finding of complex diseases and precise medicine. Using the protein quantitation assay of proteomics to find the biomarkers of complex diseases is important pipeline in the discovery of biomarkers, which has raised great interest and become a very active topic in biomedicine communities. However, it still encounters a significant challenge at small sample data. Existing high-throughput sequencing can obtain the DNA sequences with both excellent efficiency and low cost, which contributes to the discovery of sequential biomarkers from biological sequences. This project studies the sequential biomarker discovering of complex diseases and the dynamic transferring model of the sequential biomarkers. First, a reverse-index-and-pattern-growth-based algorithm is proposed to efficiently mine the closed sequential patterns. And then, this project introduces a weighted cosine similarity distance algorithm to measure the significance level of difference between any two sequence mapping pattern sets. Third, an algorithm, named Bipartite Graph-based theory is drawn to identify the sequential biomarkers and find the dynamic model from the multiple mapping pattern sets. Moreover, based on the above results, this project will release the algorithms and develop an online website for the sequential biomarker discovering, which supports the related researchers of complex diseases.

寻找和发现复杂疾病分子层面的生物标记物，有助于研究复杂疾病的发生、发展机理，对复杂疾病的早期检测、分期判断、药物靶点确定、以及实现精准医疗具有重要意义。通过蛋白质组学中的蛋白质定量技术来发现复杂疾病生物标记物是目前生物标记物发现的重要途径和研究热点，但该方法面临着小样本数据的巨大挑战。高通量测序技术的成熟，使得从生物序列中发现序列标记物具有可行性。本课题研究复杂疾病的序列标记物发现方法，以及标记物的动态迁移模型。首先提出基于倒排索引的模式增长方法高效地挖掘闭合序列模式，接着研究基于赋权向量余弦相似度方法来测度序列映射模式集合间的差异距离，进而研究基于类二分图搜索方法从多映射模式集合中发现序列标记物及其动态迁移模型。此外，根据前述研究结果，本课题拟公开算法程序源码，开发和部署序列标记物发现的 Web Server 平台，为相关研究人员提供序列分析支持。

生物标记物在临床医学领域扮演着极其重要的角色。寻找和发现癌症分子层面的生物标记物，有助于研究癌症的发生、发展机理，对癌症的前期筛选、早期检测、分期判断、药物靶点确定、以及实现精准医疗具有重要意义。通过蛋白质组学中的蛋白质定量技术来发现癌症生物标记物是目前生物标记物发现的重要途径和研究热点，但该方法面临着小样本数据以及测量误差的巨大挑战。高通量测序技术的成熟，使得从生物序列中发现序列标记物具有可行性。本课题研究癌症的敏感序列标记物发现方法，进而研究敏感序列标记物与癌症前期状态之间的因果关系，从而辅助医疗检测机构进行癌症前期筛查。首先提出基于倒排索引的前缀增长方法高效地挖掘等价类模式，接着研究基于赋权向量余弦相似度方法来测度序列映射模式集合间的差异距离，进而从多映射模式集合中发现动态迁移的序列标记物，即敏感序列标记物。此外，根据前述研究结果，本课题已开发和部署敏感序列标记物发现的Web 服务平台，该平台构建敏感序列标记物的迁移模型，给出比对序列对应个体的健康、癌前或分期状态分析报告。发现的蛋白质序列标记物为药物设计研究者提供线索。