Link Prediction And Mining Based On Disease,Gene And Drug Networks

In view of the network properties of biological and biochemical processes,such as disease phenotype,gene regulation and drug interactions,the prediction of disease-gene associations,disease-microRNA(miRNA)associations,drug-drug interactions and the corresponding concepts of Traditional Chinese Medicine(TCM)and pancreas are studied in this thesis by means of network propagation and association mining.Identification of disease-causing genes is an important part of medical research.The phenotypic similarity among diseases may reflect the interactions at the molecular level,and phenotype comparison can be used to predict disease candidate genes.Online Mendelian Inheritance in Man(OMIM)is a database of human genetic diseases and related genes that has become an authoritative source of disease.However,disease phenotypes have been described by free text in the OMIM;thus,standardization of phenotypic descriptions is needed before diseases can be compared.Several disease phenotype networks have been established using different standardization methods.Two of these networks are important for phenotypic similarity analysis:the first and most commonly used network(mimMiner)is standardized by medical subject heading,and the other network(resnikHPO)is the first to be standardized by human phenotype ontology.This paper comprehensively evaluates for the first time the accuracy of these two networks in gene prioritization with network propagation algorithm,namely,PRINCE(PRIoritizatioN and Complex Elucidation)based on protein-protein interactions using large-scale,leave-one-out cross-validation experiments.The results show that both networks can effectively prioritize disease-causing genes,and the approach that relates two diseases using a logistic function improves prioritization performance.Tanimoto,one of four methods for normalizing resnikHPO,generates a symmetric network and it performs similarly to mimMiner.Furthermore,an integration of these two networks outperforms either network alone in gene prioritization,indicating that these two disease networks are complementary.Deregulations of miRNAs are closely related to the occurrence of human disease,particularly cancer.Identifying disease-related miRNAs is essential to understand the molecular mechanisms of disease.Computational approaches can reduce the range of candidate miRNAs,which is useful to quickly discover new disease-miRNA associations.Based on functional network propagation,PMBP(Prioritizing disease MiRNA Based on PRINCE)algorithm was proposed for improving prediction of disease-miRNA associations.Results show that for the diseases without any known related miRNAs,the miRNA can be effectively predicted using disease similarities as prior information.In this case,it is impossible for random walk.And for the diseases with some known related miRNAs,PMPB achieves superior performance with the AUC values vary from 79.74%to 86.59%.In the case study of breast cancer,the associations of the predicted top 50 miRNAs with breast cancer are confirmed,indicating the validity of PMPB.For people who take multiple medications at the same time,such as the elderly or cancer patients,there may be potential interactions between different drugs,some of which can lead to serious side effects and endanger human health and even lives.Based on drug phenotypic,therapeutic,chemical,and genomic similarity,an algorithm NP(Network Propagation)is proposed for predicting drug-drug interactions.The algorithm combines drug multidimensional similarity and logistic regression preprocessing.The experiments show that the prediction using integration of multidimensional drug similarity is superior to the ones using single drug similarity.Furthmore,the drug similarity data preprocessed by the logistic regression function give better prediction results.Finally,the performance of the rank fusion and the network fusion is close to each other,and the latter performs better.The performance of NP algorithm is better than that of some existing methods,such as the traditional random walk,the recently reported structure perturbation and some frequently used machine learning methods.Symptomatic patterns of congenital syndromes may be indicative of common developmental mechanisms.A striking similarity has been discovered between the clinical features of multiple congenital anomaly/mental retardation syndromes(MCA/MRs)and the meridians and Zang-Fu theories of TCM.In the original TCM theories,there is not a proper concept for the pancreas.In this study,an effort has been made to analyze the features of MCA/MRs involving various pancreatic malformations with the TCM theories in order to obtain the pancreatic concepts in the TCM.Clinical synopses of pancreatic malformations were retrieved from the OMIM database.Following the standardization of the phenotypic features,frequently co-occurring malformations among such MCA/MRs were discovered using a frequent-pattern mining algorithm.A strong correlation has been discovered between the features of the listed MCA/MRs and the TCM concepts.The pancreas should be on the Kidney meridian,and at least in part,the pancreas has been incorporated into the TCM concept of the"Kidneys".Embryonic development of pancreas,liver and kidney are closely related.The strong correlation between the MCA/MRs and meridians suggests that the latter are genuine reflections of the connections between particular organs during embryonic development.Although the pancreas is a "missed" organ in the TCM theories,its conceptual incorporation into the kidneys is in fact very useful,which can provide new insights into the pathogenesis of pancreatic disorders and lead to more effective treatment for pancreas-related diseases.In this paper,the network propagation method is used to predict link related to disease.The mothod is improved flexibly according to the given situation and solves the specific prediction problem effectively.The improvement strategies have an important significance to solve similar problems.Finally,the specific type of disease,i.e.,pancreatic congenital syndrome,is analyzed based on the TCM by using the existing technique of data mining,and the pancreatic concepts in the TCM are discovered.