复杂疾病恶性突变的数据挖掘及预警模型 Data Mining and Early-Warning Model for the Sudden Deterioration of Complex Disease

利用基于多样本和基于单样本动态网络生物标志物法研究前列腺癌和肝癌两个时序列数据，检测疾病恶性突变的临界信号，确定动态网络生物标志物，进一步帮助医学工作者研究复杂疾病的发展变化机制，更高效准确诊断病情，及时提出合理的治疗方案。基于高通量生物分子数据，通过多样本动态网络生物标志法我们发现前列腺癌和肝癌样本分别在第6、2个时间点发生突变，与实验观测吻合，且分别有264，139个生物标记物。而实际样本数据并不完整且样本量少，此时需采用基于单样本动态网络生物标志物法检测疾病恶性突变信号，得到前列腺癌和肝癌样本分别在第6、2个时间点发生突变。最后对生物标记物进行生存分析等功能分析，发现这些标志物能较好的反映疾病临界变化信号。 Data mining and early-warning signals of prostate cancer and liver cancer are by dynamic network biomarker method based on multi-samples or single-samples. It’s vital to detect the critical point and signals of sudden deterioration, so as to diagnose the disease more accurately and put forward appropriate therapeutic plan in time. With time-course high-throughout biomolecular data, dynamic network biomarkers method based on multi-samples detected that the critical points of prostate cancer samples and liver cancer samples are the 6th time point and 2rd time point respectively, which agrees with the experiment data. In addition, 264,139 dynamical network biomakers including transcription factors were found. In fact, actual data are insufficient and the size of samples is small, and then dynamic network biomarkers method based on single-samples can be used to detect the early-warning of sudden deterioration. Also, the critical points of prostate cancer samples and liver cancer samples are the 6th time point and 2rd time point respectively based on single-samples. Finally, it shows that the found dynamic network biomakers based on multi-samples or single-samples could reflect the early-warning of sudden deterioration better after genes function analysis.