michellab
Implement of soft bond potential
Dear Sire Developers: This is an implementation of soft bond potential , including the following changes:
- Added a new perturbation type called ScalePerturbation to take care of the exclusion list changes during ring open/close. The exclusion list changes are found by biosimspace, I'll create a PR later.
- Decoupled the LJ and coulomb scale parameters from the 14 interactions, so they could be interpolated by lambda.
- Implemented the soft bond potential function for bond perturbation.
I've tested the Fxa series in the paper mentioned above, here is the result:
| pair | soft-bond run1 | soft-bond run2 | soft-bond run3 | current | exp | schrodinger |
|---|---|---|---|---|---|---|
| 4c->edo | -1.36515025 | -1.579566 | -1.599211 | -4.11 | -0.8 | -1.48 |
| 4d->edo | -1.1339635 | -0.99799 | -1.238629167 | -3.29 | -0.87 | -1.69 |
| edo->4c | 2.0459925 | 2.1593732 | 1.834021 | 5.8 | 0.8 | 1.48 |
| edo->4d | 2.04595 | 1.2482862 | 0.824815167 | 1.18 | 0.87 | 1.69 |
From the overlap matrix and pmf plot, we could see that the ring open/close is well treated:
Overlap matrix of current implementation:
PMF of current implementation:
Overlap matrix of soft-bond potential:
PMF of soft-bond potential:
Looking forward to your comments!
Hi @kexul many thanks for undertaking this significant piece of work. I will need some time to review carefully the implementation before I comment on the technical approach
It's my pleasure to contribute to this great open source project. 😄 More tests are underway, I'll post updates here.
TPSB2 series (mean and std obtained by three runs) :
| pair | soft-bond | current imp | exp | schrodinger |
|---|---|---|---|---|
| mol1->mol2 | 0.74±0.03 | 1.15±0.10 | 0.62 | 0.16 |
| mol2->mol1 | -0.81±0.07 | -1.29±0.17 | -0.62 | -0.16 |