Uus/pMeC6H4-/キ 发表于 2025-12-28 23:09
是什么算法生成的NHC过渡金属配合物的smiles字符串,以及RDKit的ETKDGv3()这个算法确认能处理此类情形的3D ...
Jiang127127 发表于 2025-12-29 08:44
谢谢回复!
smiles是通过chemdraw先画分子结构,生成smiles,再通过RDkit:
mol = Chem.MolFromSmiles
wxyhgk 发表于 2025-12-29 08:51
SMILES 本身就没有 3D 的信息,你拿 SMILES 生成 3D 结构本身就是不可行的。
当然有部分深度学习模型可 ...
Jiang127127 发表于 2025-12-29 09:36
谢谢回复!
我之前没有补充,是smiles先转化成mol,再通过mol生成构象这样的
Uus/pMeC6H4-/キ 发表于 2025-12-29 09:11
平时有ChemDraw的用户不是画完键线式直接拿Chem3D生成三维结构了么,何故绕道折腾RDkit呢,是试过发现完 ...
wzkchem5 发表于 2025-12-29 13:06
只是键长不合适,其他没有硬伤的话,产生完构象做一个GFN2-xTB预优化就行了
风起~ 发表于 2025-12-30 09:08
10.1038/s41467-023-38169-2
专门搞这个的
First, we initialize chemically-meaningful ligand geometries from the SMILES strings using the Openbabel[42] package, which are then relaxed with either MMFF94[43] or UFF[44].
With different ligand conformations at multiple sites at the metal surface, rapid methods for ranking conformations and reducing numbers of structures breaking chemical sanity are needed for complex assembly. By default we use GFN2-xTB[31] to select FF-relaxed ligand geometries by placing each ligand conformer on the complex and evaluating the total energy to select low-total-energy geometries for each ligand in order from highest to lowest denticity.
Jiang127127 发表于 2025-12-29 22:05
谢谢老师的指导!按照您的方法问题解决啦,想请教您:我尝试了xtb三种优化方法,gfnff就只是让CuCl键长合 ...
| 欢迎光临 计算化学公社 (http://bbs.keinsci.com/) | Powered by Discuz! X3.3 |