Molecules with rotatable bonds can exist in multiple conformations, and locating the lowest energy conformation is crucial for ensuring the accuracy of subsequent tasks. Running a conformational search can also reveal the conformational preferences of different bonds, and show how strained various alternate conformations are relative to the global minimum. Rowan's conformational searching workflow uses fast low-level methods to explore conformational space, followed by more accurate machine-learned interatomic potentials to score the poses generated.
Conformational searching in Rowan proceeds in two stages: conformer generation and conformer optimization. Conformer generation employs either RDKit (for reckless and rapid modes) or CREST (for careful and meticulous modes).
RDKit-based methods use the ETKDG v2 algorithm to generate a large number of initial conformers (200 for reckless, 300 for rapid). Each generated conformer is then optimized and scored using the MMFF94 forcefield, and the generated conformer ensemble is screened using CREST's cregen utility with an energy cutoff of 50 kcal/mol. The generated conformer ensemble is then re-screened using CREST's screen utility, with a 2% rotational constant threshold and a 0.25 Ã… RMSD cutoff.
CREST-based methods use the iMTD-GC algorithm with rotamer metadynamics. Conformer searches run in "careful" mode employ 6 metadynamics runs with a run scaling factor of 0.5 and an energy window of 5.0 kcal/mol (CREST --quick mode), while conformer searches run in "meticulous" mode employ 14 metadynamics runs with a run scaling factor of 1.0 and an energy window of 6.0 kcal/mol. Screening and filtering are performed automatically by CREST.
Each member of the generated conformer ensemble is then optimized using the user's specified methodology, with solvent corrections applied as applicable.
Rowan's conformational searching workflow supports transition state (TS) conformer searches. A transition state (TS) conformer search should be performed on fully optimized TS, not a guess TS.
Transition states require extra care, as conformer generation schemes will natively relax the TS to a local minimum. User provided constraints are required to ensure that the conformer generation creates conformers that look like the TS. Once the conformers have been generated, further TS optimizations are performed without constraints.
Rowan's conformer workflow can be run in four modes: meticulous, careful, rapid, or reckless. Here's what selecting each mode tunes:
| Mode | Meticulous | Careful | Rapid | Reckless |
|---|---|---|---|---|
| program used for conformer generation | CREST | CREST | RDKit | RDKit |
| level of theory used for conformer generation | GFN2-xTB | GFN-FF | MMFF | MMFF |
| mode (CREST) | normal | quick | ||
| number of initial conformations (ETKDG) | 300 | 200 | ||
| initial energy cutoff (kcal/mol) | 15 | 10 | 15 | 10 |
| RMSD similarity cutoff (Ã…) | 0.00 | 0.00 | 0.10 | 0.25 |
| max number of conformers (xTB) | 500 | 150 | 100 | 50 |
| final energy cutoff (kcal/mol) | 10 | 5 | 5 | 5 |
| max number of conformers (AIMNet2) | 150 | 50 | 50 | 20 |
In general, the "rapid" mode provides a good balance between speed and accuracy for most applications.