A quantitative structure–activity relationship (QSAR) model was built from a dataset of 54 peptide-type compounds as SARS-CoV inhibitors. The analysis was executed to identify prominent and hidden structural features that govern anti-SARS-CoV activity. The QSAR model was derived from the genetic algorithm–multi-linear regression (GA-MLR) methodology. This resulted in the generation of a statistically robust and highly predictive model. In addition, it satisfied the OECD principles for QSAR validation. The model was validated thoroughly and fulfilled the threshold values of a battery of statistical parameters (e.g. r ² = 0.87, Q ² _loo = 0.82). The derived model is successful in identifying many atom-pairs as important structural features that govern the anti-SARS-CoV activity of peptide-type compounds. The newly developed model has a good balance of descriptive and statistical approaches. Consequently, the present work is useful for future modifications of peptide-type compounds for SARS-CoV and SARS-CoV-2 activity.

从54种作为SARS-CoV抑制剂的肽类化合物的数据集中建立了定量构效关系（QSAR）模型。执行该分析以识别控制抗SARS冠状病毒活动的突出和隐藏的结构特征。QSAR模型源自遗传算法-多线性回归（GA-MLR）方法。这导致生成了统计上可靠且具有高度预测性的模型。此外，它满足了OECD关于QSAR验证的原则。对该模型进行了充分验证，并满足了一系列统计参数的阈值（例如r ²  = 0.87，Q ² _loo = 0.82）。派生模型成功地确定了许多原子对作为控制肽类化合物抗SARS-CoV活性的重要结构特征。新开发的模型在描述和统计方法之间取得了很好的平衡。因此，当前的工作对于将来针对SARS-CoV和SARS-CoV-2活性的肽型化合物的修饰是有用的。