crest-lab / crest

CREST - A program for the automated exploration of low-energy molecular chemical space.
https://crest-lab.github.io/crest-docs/
GNU Lesser General Public License v3.0
182 stars 42 forks source link

Initial geometry optimization failed #313

Open Jevon-Du opened 4 days ago

Jevon-Du commented 4 days ago

Hello,

I am trying to generate conformations for a cyclic peptide using crest v3.0.1, precompiled binary, commit (1782d7d), xtb version 6.6.1. I get an error when using this command crest xtbtopo.mol --gfn2 -g h2o -T 14, but the program runs fine when using command crest xtbtopo.mol --gfn2 -g chcl3 -T 14. I don't know why. And I tried to use the crest version 2.12(Compatible with xTB version 6.4.0) and it worked fine.

The last two lines of stdout of command crest xtbtopo.mol --gfn2 -g h2o -T 14 is

  Initial geometry optimization failed!
  Please check your input. 

test.pdb is the original cyclic peptide structure. xtbopt.pdb and xtbtopo.mol are the cyclic peptide structures after geometry optimization using xtb 6.6.1. The geometry optimization commond is xtb test.pdb --opt. xtbtopo.mol was used as input file of crest v3.0.1. A xyz format of xtbtopo.mol was also used as input file, but I still got the same error as above.

PS: Please change the file suffix txt to the correct format, test.txt to test.pdb, xtbopt.txt to xtbopt.pdb, xtbtopo.txt to xtbtopo.mol.

pprcht commented 2 days ago

Hi, since you get different performance for two different implicit solvents it quite possible is because the implicit solvation implementation in tblite is currently inconsistent with xtb, as mentioned e.g. here #312 There is currently not too much I can do from this side, but I spoke to the tblite developers about it. You could try using --alpb water. This may work, but be aware that the inconsistency is also present in this case.

Jevon-Du commented 7 hours ago

Hi,

Thank you for your reply. I tried using --alpb water(crest v3.0.1), but it doesn't work.
OK, I will use crest v2.12 to complete my work.

Thanks again.