Combinatorial preferences affect molecular similarity/diversity calculations using binary fingerprints and Tanimoto coefficients
Combinatorial preferences affect molecular similarity/diversity calculations using binary fingerprints and Tanimoto coefficients
复制标题
DOI：
10.1021/ci990316u
复制
发表时间：
2000-01-01
期刊：
影响因子：
--
通讯作者：
Bajorath, J
中科院分区：
文献类型：
--
作者：
Godden, JW;Xue, L;Bajorath, J
发展了一种组合方法来计算指定长度的二进制指纹(FP)表示的TAnimoto系数(TC)的完全分布，而不考虑它们所反映的化学参数。计算了由多达67个比特位置组成的FP的理论TC分布，这揭示了某些Tc值的显著统计偏好。使用不同的FP计算了一个大型化合物数据库中的Tc分布，反映了我们一般分析确定的影响。在这些发现的基础上，平均T_c受到统计上的偏好值的影响。
A combinatorial method was developed to calculate complete distributions of the Tanimoto coefficient (Tc) for binary fingerprint (FP) representations of specified length, regardless of the chemical parameters they reflect. Theoretical Tc distributions were calculated for FPs consisting of up to 67 bit positions which revealed significant statistical preferences of certain Tc values. Calculation of Tc distributions in a large compound database using different FPs mirrored the effects identified by our general analysis. On the basis of these findings, an average Tc is biased by statistically preferred values.