Two structures of the same protein can show different conformations, but knowing the starting and ending shapes does not reveal how the protein might move between them. For molecular modelers, bridging that gap is a practical challenge: a straight interpolation may not produce a useful motion, and inspecting endpoints alone can hide the range of possible transitions.
SAMSON’s Protein Path Finder offers a way to search for candidate transition paths between two conformations. It uses ART-RRT: T-RRT guides the pathway search, while ARAP generates motion between states. Constrained minimization is applied as new states are generated, helping keep the explored motions physically reasonable. The result is a set of candidate paths to inspect—not a single guaranteed account of how the protein moves.
Start with comparable structures
A useful search begins with two conformations of the same system. The structures need matching chains, residues, and atoms; otherwise, the app cannot treat them as alternate conformations of one protein. If your structures are in separate files, prepare them individually and combine them as separate models in one PDB file. For example:
|
1 2 3 4 5 6 7 |
MODEL 1 ... ENDMDL MODEL 2 ... ENDMDL END |
Before searching with your own model, check for alternate locations, unwanted ligands, solvent, or ions, and add hydrogens as needed. Missing residues or heavy atoms may require repair first. The supplied Adenylate Kinase sample is already prepared and includes the 4AKE and 1AKE conformations, making it a convenient way to learn the workflow before adapting it to another protein.
Make the motion controllable
One important setup choice is which atoms actively guide the motion. In ARAP, active atoms control the protein’s movement; the remaining atoms are passive and follow their motion. Choosing a small, meaningful set can make the setup easier to reason about. In the tutorial’s example, two backbone alpha-carbon atoms—one in GLY 12 and one in ARG 123—are selected as active atoms. The app can use selected nodes from the document, and the tutorial’s sample includes a group containing those atoms.
The sampling box also matters. It defines the region in which active atoms are sampled, so its size and position influence the motions the search explores. The app initially creates a box enclosing the protein in both endpoint conformations; the tutorial then demonstrates setting a 200 Å cube. Treat the box as part of your search setup, not merely as a display element.
Set up energy evaluation before searching
The system should already be minimized before the planner runs. The tutorial uses Universal Force Field (UFF) as the interaction model and FIRE as the state updater, with a step size of 1 fs and one step. If FIRE is not available in the app, install the required state updater first. When UFF asks whether to use existing bonds, the tutorial specifies using them.
These settings establish the energy evaluation and minimization used during the search. The tutorial’s example uses 20 ARAP-modeling iterations and 20 constrained minimization iterations for each generated state. Its search parameters also include two runs, a 1 Å RRT extension step size, and T-RRT temperature settings. These are example values to reproduce the tutorial workflow, not universal settings for every protein or research question.
Read the results as candidates
As paths are found, they appear in the Results tab. Compare the number of states, minimum and maximum energies, saddle, barrier, and elapsed search time. Selecting a path lets you move through its conformations and inspect the corresponding energy curve. Multiple paths can be selected to compare their curves, which is useful when the search finds more than one candidate route.
A fixed seed makes a run reproducible under the same setup; without it, the planner uses a random seed. The alignment strategy can make searching faster, although the documentation notes that it may produce artifacts in rare cases. These details are worth recording when comparing runs: a path depends on both the model and the choices used to search it.
Paths can be exported as trajectories, and conformations can be exported at a chosen interval. For further refinement, the documentation points to the P-NEB method. This gives a practical workflow: verify the endpoint structures, choose active atoms and a sampling region, run and compare candidate paths, then export or refine the results as appropriate.
For setup details and the complete walkthrough, see the Protein Path Finder tutorial.
Note: SAMSON and all SAMSON Extensions are free for non-commercial use. Get SAMSON.