New Benchmark for Chemical Nomenclature Software

We propose a new, robust benchmark, called Percentage
Round Tripping of Canonical Isomeric SMILES (%RTCS), for assessing
the ability of chemical nomenclature software to convert chemical
structures to names and chemical names to structures. The benchmark
is based on a string comparison between canonical isomeric SMILES
generated from the original structure and the resultant structure
from round tripping. Using the latest version of the OpenEye chemical
nomenclature toolkit, Lexichem v2.1.0, we report %RTCS values of over
92% on average for a variety of challenging compound collections.

我们提出了一款全新且稳健的基准测试工具，命名为百分比往返标准化异构SMILES（Percentage Round Tripping of Canonical Isomeric SMILES，%RTCS），用于评估化学命名软件实现化学结构与化学名称互相转换的能力。该基准测试通过字符串比对的方式，对比原始结构生成的标准化异构SMILES，与经往返转换后所得结构对应的标准化异构SMILES。本研究采用最新版OpenEye化学命名工具包Lexichem v2.1.0，针对多种极具挑战性的化合物集合开展测试，测得%RTCS的平均数值超过92%。