Two structures can show where a protein starts and ends, but they do not tell you how it moves between those conformations. For molecular modelers, that gap can make it difficult to explore a proposed transition, inspect intermediate structures, or prepare a path for further analysis.
SAMSON’s Protein Path Finder addresses this by searching for possible transition paths between two conformations of the same protein. It uses ART-RRT, combining T-RRT for pathway search with ARAP for motion generation, and applies constrained minimization as new states are generated. The result is a set of candidate paths to inspect—not, by itself, proof that a protein follows a particular biological route.
Start with compatible structures
A useful path search depends on having a suitable pair of endpoints. The tutorial’s sample document contains adenylate kinase chain A in the 4AKE and 1AKE conformations, labeled start and goal. It is a convenient way to learn the workflow before bringing in your own system.
For your own structures, prepare both conformations and check that they describe the same system: the chains, residues, and atoms need to correspond. The tutorial recommends removing alternate locations, ligands, solvent, and ions, and adding hydrogens. If residues or heavy atoms are missing, the PDBFixer extension can help repair the protein. When the conformations come from separate files, prepare them first and then combine them as separate models in a single PDB file.
Before searching, minimize the system using Edit > Minimize. The documented workflow uses Universal Force Field (UFF) for minimization and energy evaluation, together with the FIRE state updater. In Protein Path Finder’s Settings tab, select UFF and FIRE; if prompted, apply the model and use the existing bonds. Set FIRE’s step size to 1 fs and its number of steps to 1, as shown in the tutorial.
Choose what can guide the motion
After opening Set up the system, retrieve the conformations from the active document and assign the start and goal. Then select the active ARAP atoms. These atoms control the protein motion; the remaining atoms follow their movement according to the ARAP method.
This choice is important because it shapes the motion the search can explore. In the sample, two backbone alpha-carbon atoms are used: the CA atoms in GLY 12 and ARG 123. Select the corresponding group in the Document view, then click Add in the app. Check the active-atom count in Advanced information, and use the Select buttons to verify which atoms were assigned. If the selection is not what you intended, reset it and choose again.
Set the search region and make runs reproducible
The sampling box defines where the active atoms can be sampled. Its size and position therefore influence the pathways the search can find. The tutorial starts with a box enclosing the protein in both endpoint conformations and demonstrates setting a cube 200 angstroms along each dimension. Treat this as a tutorial setting, not a universal value: inspect the visualization and choose a region appropriate to the motion and structure you are studying.
Search settings also affect the process. The example uses two runs, 20 ARAP-modeling iterations, and 20 constrained-minimization iterations per generated state. It sets the initial temperature to 0.001 K, the temperature factor to 2, failures before increasing temperature to 1, and the RRT extension step size to 1 A. A maximum elapsed time can limit each run.
If you need to reproduce a search, enable Use seed and record the seed value and settings. With this option enabled, the seed increments by one after each run. Without it, the app uses a random seed. The alignment strategy can make the search faster, according to the documentation, but may produce artifacts in rare cases; compare results with care if you use it.
Inspect paths instead of judging them by one number
Click Run to search. The Advanced information panel reports elapsed time, tree nodes, run number, and paths found. Found paths appear in the Results tab. The table includes the number of states, minimum and maximum energies, saddle and barrier values, time, and endpoint energies.
Select a path to view its energy curve, then move the slider to examine individual conformations in the viewport and their UFF energies. You can select multiple paths to compare their energy curves. These values help describe the computed paths, but they are not a substitute for evaluating whether a path is physically or biologically plausible. Review the structures and motion as well as the table, and keep the force field and search setup in mind when interpreting the energies.
When you want to continue analysis, export selected paths as trajectories or export conformations at a chosen interval. For refinement, the documentation points to the P-NEB app, which can improve resulting paths using the parallel Nudged Elastic Band method.
This workflow is most useful when you treat path finding as a way to generate and inspect candidate transitions: prepare matching endpoints, make deliberate choices about active atoms and sampling space, record settings when reproducibility matters, and review the resulting motion before drawing conclusions.
Learn more in the Protein Path Finder tutorial.
Note: SAMSON and all SAMSON Extensions are free for non-commercial use. Get SAMSON.