| Drug targets are defined as biological macromolecules with pharmacodynamic functions in vivo.The prediction of drug potential targets is very important for the early drug molecule research and development,safety evaluation,new drug use,clarify the mechanism of natural products and so on.Target prediction and validation as an important part of innovative drug discovery,for innovative drug discovery and research is of great significance.The use of computational tools and tools to discover and predict potential targets for drugs can provide important guidance and supporting evidence for drug experiments.In this paper,we mainly study various methods of predicting drug targets by computer technology,and by means of machine learning and data mining,we use the structure information combined with the biology data such as gene-protein-compound interaction,And the new clustering algorithm is used to carry out molecular clustering in the target to improve the prediction ability.Mainly in the following aspects:1.A new clustering method is used in the target molecular processing.The main idea is that the cluster centers are surrounded by neighbors with low local density and have a relatively large distance from any point with higher density,Without the need to set the number of classes,the calculation speed,the distribution of different types of data for the advantages of good clustering.2.The models for predicting drug targets by means of machine learning and data mining are modeled,and the acquisition and preliminary treatment of drug targets with active interactions are carried out.By comparing the structural similarity of pharmacophore and chemical structure,we can use the structure information of known drug target and active compound to predict the target,and our data processing method is based on high-performance computer and has faster processing speed.3.SHAFTS by using three dimensional similarity algorithm,through small molecular similarity classification and target binding site,to build a target prediction model based on three dimensional similarity,needle new similarity clustering method and the description of small molecules.At the same time,the model and the algorithm are tested in the tianhe-2,and the research progress is accelerated and its application is more widely. |