唑类化合物密度的全息定量构效关系模型研究

Hologram Quantitative Structure-Activity Relationship Study on the Density of Azole-based Energetic Compounds

唑类化合物是一种爆轰性能与密度相关的重要的能量材料,建立了唑类化合物的密度的全息定量构效关系(HQSAR)模型。模型参数选择为fragment size:“1~2”,fragment distinction:“A、B、H、DA”可建立最佳HQSAR模型。采用两种常规验证技术,即外部测试集验证和留一交叉验证(LOO-CV)来评估HQSAR模型的预测精度。采用外部测试集验证和留一交叉验证两种方法对所建立模型进行了验证,外部测试集验证中的RMSRE、RMSEP、MRE、Q_(2)_(F3)、r^(2)_(m)和Δr^(2)_(m)分别为0.020、0.044、2.04%、0.891、0.670、0.220。留一交叉验证中的RMSRE、RMSECV、MRE、r^(2)_(m)和Δr^(2)_(m)分别为0.043、0.070、3.20%、0.602和0.237。结果表明,所建立的HQSAR模型可用于预测唑类化合物的密度。

Azole-based compounds are a kind of important energy material.Their detonation performance significantly relates to their density.This work investigated the quantitative relationship between the density and structure of azole-based compounds through hologram quantitative structure-activity relationship(HQSAR)technique.The fragment parameters of the established optimal HQSAR model were fragment size:“1~2”,fragment distinctions:“A,B,C,H,DA”.Two conventional validation techniques,external test set validation and leave-one-out cross-validation(LOO-CV),were utilized to evaluate the prediction accuracy of the HQSAR model.Root mean square relative error(RMSRE),root mean square error of prediction(RMSRP),mean relative error(MRE),concordance correlation coefficient(CCC),Q_(2)_(F3),r^(2)_(m) andΔr^(2)_(m) of external test set validation are 0.020、0.044、2.04%、0.891、0.670、0.220,respectively.RMSRE,root mean square error of cross validation,MRE,r^(2)_(m) andΔr^(2)_(m) of LOO-CV are 0.043、0.070、3.20%、0.602 and 0.237,respectively.It is demonstrated that the established HQSAR model is available for predicting the density of azole-based compounds.