FOLDIFF · DYNAMICS COMPARISON
What changes after a mutation?
Compare observed configurations of WT and variants. Matched coordinates, shared scales and explicit condition checks.
This stage analyzes the selected residue’s φ / ψ angles in existing MD trajectories. It describes local backbone motion, not a full protein landscape or absolute mutation ΔΔG.
01 · DATA
From topology to comparison
Import your topology and trajectories
Prepare a larger dataset locally ↓ · File processing and limits
02 · COMPARISON
Two distributions, one coordinate system
WT and variant share a color scale. Energy-map zeros are chosen separately; their difference is not calculated.
Variant
Frequency difference
Gray means no observations. Hatching marks low counts in at least one compared bin. On the difference map, teal favors WT and terracotta favors the variant; units are percentage points. Selection is linked across all three maps.
Arrow keys move across the map. Means shown below are circular; they do not describe the full distribution.
Two actual frames
WT · variant. Fragment of up to five neighboring residues on each side; aligned on shared Cα atoms. Frames from the selected region, not a transition pathway.
03 · IS THE SAMPLING SUFFICIENT
Separate observation from inference
Metadata, preparation and limitations
No autosave. Projects contain both datasets and parameters; reopening recomputes the result.
04 · WHAT CAN BE CONCLUDED
Compare motion without claiming function
We match protein, construct, force field, solvent, ensemble, temperature and context using supplied metadata. Sequences use identical residue numbering: substitutions only, no indels. Matching metadata does not validate the simulation physics.
Probabilities are estimated separately; the difference is P(variant) − P(WT). JSD and overlap describe distributions. They do not determine mutation success, causality, activity, transition rates or absolute ΔΔG. Unknown temperature permits dimensionless F/RT only, not kJ/mol.
Published example: DHFR WT and L28R with TMP⁺ and NADPH, CHARMM36/TIP3P. Five files per condition continue one microsecond trajectory. Boundary duplicates are checked; files are not counted as independent replicates. This validates our analysis of published MD, not every conclusion of the paper.
Original trajectories · CC BY 4.0 · Cetin et al., 2023 · Methods & formats · Validation record