论文部分内容阅读
针对新化合物作用靶点的大规模识别问题,提出基于欧式距离的化合物作用靶点识别方法。本方法以作用于同一靶点的药物分子结构为基础,计算其分子结构描述符,并以结构描述符为基础计算药物两两之间的欧式距离,利用置信区间估计的方法估算作用于该靶点化合物间的距离上限。若新化合物与上述各药物分子之间的距离小于估算的距离上限,则判断新化合物作用于该靶点。利用已知作用关系的177个靶点、767个化合物的数据对所建立的方法进行检验,判断总正确率为70.66%,对于有活性的化合物判断正确率达到82.20%,结果证明该方法可以用于对新化合物生物活性的大规模识别,对设计活性筛选实验具有参考价值。
In view of the large-scale identification of the new compounds target, a method of target recognition based on European distance was proposed. Based on the molecular structure of the drug acting on the same target, the molecular structure descriptor is calculated and the Euclidean distance between the two drugs is calculated on the basis of the structure descriptor. The method uses the method of confidence interval estimation to estimate the effect on the target The maximum distance between compounds. If the distance between the new compound and each of the above drug molecules is less than the upper limit of the estimated distance, it is judged that the new compound acts on the target. The established method was tested by using the data of 177 target compounds and 177 compounds with known interaction. The accuracy of the method was 70.66% and that of the active compound was 82.20%. The results showed that this method could be used The large-scale identification of biological activity of new compounds is of reference value for the design of active screening experiments.