Zanardi María Marta, Sortino Maximiliano A, Sarotti Ariel M
Facultad de Química e Ingeniería del Rosario, Pontificia Universidad Católica Argentina, Av. Pellegrini 3314, Rosario, 2000, Argentina; Área Farmacognosia, Facultad de Ciencias Bioquímicas y Farmacéuticas, Universidad Nacional de Rosario, Suipacha 531, Rosario, 2000, Argentina.
Área Farmacognosia, Facultad de Ciencias Bioquímicas y Farmacéuticas, Universidad Nacional de Rosario, Suipacha 531, Rosario, 2000, Argentina.
Carbohydr Res. 2019 Feb 15;474:72-79. doi: 10.1016/j.carres.2019.01.011. Epub 2019 Jan 28.
Hyacinthacines are important members of the pyrrolizidine family, with several compounds having ambiguous, revised or unverified structures. Herein we thoroughly explored the performance DP4 and DP4+ for the in silico stereoassignment of hyacinthacines A1, A2 and five synthetic isomers. The results suggested that the quality of the predictions strongly depended on the conformational landscape provided by DFT energies, with five compounds correctly assigned. In the two cases incorrectly classified we found that the source of the problem was conformational in nature, with spurious conformations being considerably over-stabilized by intramolecular H-bondings. We showed that neglecting such shapes resulted in a noteworthy improvement, with all compounds correctly assigned in high confidence (>99.9%).
风信子碱是吡咯里西啶家族的重要成员，有几种化合物的结构不明确、已修正或未经证实。在此，我们全面探讨了DP4和DP4 +算法用于风信子碱A1、A2及五种合成异构体的计算机辅助立体构型确定的性能。结果表明，预测质量很大程度上取决于密度泛函理论（DFT）能量提供的构象态势，有五种化合物被正确构型。在两个分类错误的案例中，我们发现问题根源本质上是构象方面的，虚假构象因分子内氢键作用而过度稳定。我们表明，忽略这些构象会带来显著改进，所有化合物都能以高置信度（>99.9%）正确构型。
