|
|
收个尾
之所以出现这种和C相连的H距离过近,是因为AutoDock-GPU采用了联合原子力场,将H的参数整合到了父原子中。最后利用Meeko得到的pdb,其中C-H是后来被添加,但是是根据几何规则添加的,并不严格。(之前扫了一眼pdbqt,看见有三个命名相似的H原子,以为是C上的H,当时没有想这么多,这其中是N端的3个氢,C-H的H确实是默认整合到父原子上了)因此我认为单纯的对接后可视化没有意义,至少动力学模拟后才可以用aNCL之类的
原文“Since hydrogens bonded to carbon are excluded by default in AutoDock, exported positions of these hydrogens are calculated by RDKit. This can be annoying if a careful forcefield minimization is employed before docking, as probably rigorous Hs positions will be replaced by the RDKit geometry rules, which are empirical and much simpler than most force fields.”。链接“https://meeko.readthedocs.io/en/release-doc/export_usage.html”
其次AutoDock和AutoDock-GPU打分函数应当是一致的,可以相互比较,各种流程是一致的,除了可以用GPU加速和预处理软件不同
原文“AutoDock-GPU is the fastest full-service docking engine available at RSD3. It is a version of AutoDock4.2.6 accelerated by OpenCL and Cuda. It leverages the embarrassingly-parallelizable Lamarckian Genetic Algorithm of AutoDock by processing ligand-receptor poses in parallel over multiple compute units.”。链接“https://rsd3.scripps.edu/services/”
最后 Moving Ligand-Fixed Receptor和 Moving Ligand-Moving Receptor 还是没有弄懂,客观的现象就是Moving Ligand-Fixed Receptor=(vdW + Hbond + desolv Energy)+(Electrostatic Energy ), Moving Ligand-Moving Receptor出现在柔性对接中;因此目前理解是在分子对接中弱相互作用力占主导,而柔性对接中的弱相互作用力占比却很低,暂且不理解;因此最终的解决方案就是采用半柔性对接后直接筛选跳过柔性对接,然后进行动力学模拟 |
|