# Binding free energy calculation

Score a GROMACS protein-ligand trajectory with gmx_MMPBSA, check the run for stability, and read MM/GBSA or MM/PBSA totals with per-residue contributions before deciding which analogues deserve FEP or an assay.

[Run gmx_MMPBSA](https://proteiniq.io/app/gmx-mmpbsa)[Open end-point workflow](https://proteiniq.io/app/workflows/templates?template=binding-free-energy-end-point)

1. ## Find a starting pose

   An end-point calculation is only as good as the complex it starts from. Dock with Vina or GNINA, or cofold with Boltz-2, then screen the poses for clashes and broken geometry before simulating.

   - [![AutoDock Vina](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fautodock-vina-announcement.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### AutoDock Vina Docks ligands to produce the starting poses for simulation.](https://proteiniq.io/app/autodock-vina)
   - [![GNINA](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fgnina.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### GNINA Docks and rescores poses with CNN scoring as a second opinion before MD.](https://proteiniq.io/app/gnina)
   - [![Boltz-2](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fboltz-2.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### Boltz-2 Predicts a complex and an affinity value without simulation, for fast triage.](https://proteiniq.io/app/boltz-2)
   - [![PoseBusters](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fposebusters.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### PoseBusters Checks docked poses for chemical and geometric errors before they are simulated.](https://proteiniq.io/app/posebusters)
2. ## Set protonation states

   Charge states change the electrostatics in every later step. Dimorphite-DL enumerates ligand states at your pH, and PROPKA estimates pKa values for protein residues. In the HIV-1 protease example, Asp25 comes out at 4.35 in chain A and 7.89 in chain B.

   ![PROPKA on ProteinIQ with HIV-1 protease from PDB 1HSG, a table of residue pKa values including Asp25 at 4.35 in chain A and 7.89 in chain B](https://proteiniq.io/_next/image?url=%2Fimages%2Fuse-cases%2Fscreens%2Fpropka-hiv-protease-mk1-light.webp&w=1920&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh)![PROPKA on ProteinIQ with HIV-1 protease from PDB 1HSG, a table of residue pKa values including Asp25 at 4.35 in chain A and 7.89 in chain B](https://proteiniq.io/_next/image?url=%2Fimages%2Fuse-cases%2Fscreens%2Fpropka-hiv-protease-mk1-dark.webp&w=1920&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh)

   [Open the HIV-1 protease job](https://proteiniq.io/app/propka?jobId=cfbi6jmxb5bgoaergbdvvr0z)

   - [![Dimorphite-DL](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fligand-protonation.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### Dimorphite-DL Enumerates ligand protonation states at a chosen pH before parameterization.](https://proteiniq.io/app/dimorphite-dl)
   - [![PROPKA 3](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fpropka-3.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### PROPKA 3 Estimates pKa values of binding-site residues to choose their protonation states.](https://proteiniq.io/app/propka)
3. ## Simulate and check the trajectory

   Run the prepared complex in GROMACS or OpenMM, then read RMSD and RMSF before scoring any frames. A ligand that leaves the pocket makes the later numbers meaningless. The OpenMM example is a 1 ns run of ubiquitin from PDB 1UBQ.

   ![OpenMM on ProteinIQ with ubiquitin from PDB 1UBQ set up for a 1 ns molecular dynamics run, with a 20-model trajectory in the viewer](https://proteiniq.io/_next/image?url=%2Fimages%2Fuse-cases%2Fscreens%2Fopenmm-ubiquitin-1ns-light.webp&w=1920&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh)![OpenMM on ProteinIQ with ubiquitin from PDB 1UBQ set up for a 1 ns molecular dynamics run, with a 20-model trajectory in the viewer](https://proteiniq.io/_next/image?url=%2Fimages%2Fuse-cases%2Fscreens%2Fopenmm-ubiquitin-1ns-dark.webp&w=1920&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh)

   [Open the OpenMM job](https://proteiniq.io/app/openmm?jobId=qnljs9lrsjnf1bokt7vkw8tq)

   - [![GROMACS](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fgmx-mmpbsa.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### GROMACS Runs a prepared protein-ligand system from a native TPR or file ZIP and returns the trajectory gmx_MMPBSA reads.](https://proteiniq.io/app/gromacs)
   - [![OpenMM](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fopenmm.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### OpenMM Simulates a protein with a positioned SDF or MOL2 ligand using OpenFF Sage parameters.](https://proteiniq.io/app/openmm)
   - [![MD Trajectory Analysis](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fmd-trajectory-analysis.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### MD Trajectory Analysis Checks a trajectory with RMSD, RMSF and distance metrics before frames are scored.](https://proteiniq.io/app/md-trajectory-analysis)
4. ## Score the frames

   gmx_MMPBSA reads the GROMACS trajectory and returns MM/GBSA or MM/PBSA totals with per-residue contributions. These are end-point estimates for ranking analogues, not measured affinities. ProLIF lists which residues a pose contacts.

   - [![gmx_MMPBSA](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fgmx-mmpbsa.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### gmx_MMPBSA Calculates MM/GBSA and MM/PBSA binding free energies from GROMACS trajectories, with per-residue decomposition.](https://proteiniq.io/app/gmx-mmpbsa)
   - [![ProLIF](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fprolif.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### ProLIF Lists protein-ligand interactions in a pose or trajectory frame.](https://proteiniq.io/app/prolif)
5. ## Run a hydration free energy

   OpenFE returns the absolute hydration free energy of one neutral ligand. It is not a binding free energy, and relative or absolute binding free energy calculations are not hosted tools on ProteinIQ.

   - [![OpenFE](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fopenfe.webp&w=96&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) ### OpenFE Calculates the absolute hydration free energy of one neutral ligand with OpenFE alchemical sampling.](https://proteiniq.io/app/openfe)

## More small-molecule discovery workflows

## What is binding free energy calculation?

Binding free energy calculation is a family of physics-based methods that estimate how strongly a ligand binds its target, as the free energy change when the two associate in solution. A more negative value means tighter binding, and at 298 K each 1.36 kcal/mol corresponds to a tenfold change in the dissociation constant.

Two families answer the question. Alchemical methods such as [free energy perturbation](https://proteiniq.io/guides/free-energy-perturbation) transform one ligand into another through nonphysical intermediates, and rank a congeneric series most accurately. End-point methods score snapshots of the bound complex instead, for far less compute per ligand. ProteinIQ hosts the end-point route and one hydration-only alchemical protocol, not relative or absolute binding calculations.

Neither family gives a trustworthy absolute affinity by default. End-point totals are usually much larger in magnitude than measured values and should only rank ligands inside one protocol, which the [MM/PBSA and MM/GBSA comparison](https://proteiniq.io/guides/mm-pbsa-vs-mm-gbsa) sets out. Every result inherits the starting pose, so a wrong binding mode returns a precise answer to the wrong question.

## Which binding free energy tool to use

| What you want to do | Tool | What you get back |
| --- | --- | --- |
| Score a simulated complex with MM/GBSA or MM/PBSA | [gmx_MMPBSA](https://proteiniq.io/app/gmx-mmpbsa) | `DELTA TOTAL` in kcal/mol with energy components and per-residue decomposition |
| Produce the trajectory the scoring reads | [GROMACS](https://proteiniq.io/app/gromacs) | Trajectory, energies, topology and the final structure |
| Check the complex held together before scoring | [MD Trajectory Analysis](https://proteiniq.io/app/md-trajectory-analysis) | RMSD, RMSF and distance traces per frame |
| Test the force field and charges for a ligand in water | [OpenFE](https://proteiniq.io/app/openfe) | Hydration free energy in kcal/mol with its uncertainty and MBAR overlap |
| Get a starting pose for the simulation | [AutoDock Vina](https://proteiniq.io/app/autodock-vina) | Ranked poses with a Vina score in kcal/mol |
| Triage a complex without simulating it | [Boltz-2](https://proteiniq.io/app/boltz-2) | A predicted complex with a binding probability and a predicted log10(IC50) |

Each tool runs online from structures or a prepared system and quotes its credit cost before it starts. AutoDock Vina jobs start at 15 credits, GROMACS at 100, and an OpenFE production run with three repeats is quoted at 8,400 credits.

## What goes in and what comes out

### You provide

- Complex run files

  TPR, XTC, TRR, PDB

  The GROMACS run input of the bound complex and one to four trajectories of it, concatenated in order. A TPR is recommended over a PDB.
- Index groups

  NDX

  A GROMACS index file whose receptor and ligand group numbers you enter in the settings.
- Ligand parameters

  MOL2, TOP

  An Antechamber MOL2 for a small-molecule ligand, unless a complex topology already supplies its parameters.

### You get back

- Binding free energy table

  CSV

  DELTA TOTAL in kcal/mol with its van der Waals, electrostatic, polar and nonpolar solvation components for each requested method.
- Per-frame and per-residue energies

  CSV

  Energies for every analyzed frame, and residue contributions when decomposition is on.
- Native result files

  MMXSA, IN, LOG

  The gmx_MMPBSA report, the effective input file, run log and COMPACT_MMXSA_RESULTS.mmxsa for the gmx_MMPBSA analyzer.
- Trajectory checks

  CSV

  RMSD, RMSF and distance traces from MD Trajectory Analysis for choosing which frames to score.

## Know what each method can tell you

Background guides on the methods, scores, and numbers behind these workflows.

- [![The three end states of an MM/PBSA calculation, avidin with biotin bound, avidin alone and biotin alone, each surrounded by continuum solvent on a grid.](https://proteiniq.io/_next/image?url=%2Fimages%2Fguides%2Fmm-pbsa-vs-mm-gbsa%2Ffeatured.webp&w=1920&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) Docking · 17 min ### MM/PBSA vs MM/GBSA: how end-point binding free energies work MM/PBSA and MM/GBSA estimate protein-ligand binding free energy from molecular dynamics snapshots. Learn the energy terms, how Poisson-Boltzmann and generalized Born solvation differ, single- versus three-trajectory protocols, the entropy term, and what benchmark studies say about accuracy.](https://proteiniq.io/guides/mm-pbsa-vs-mm-gbsa)
- [![Thermodynamic cycle for a hydrogen-to-chlorine change on phenol: the ligand in a protein pocket on top and in water below, joined by rows of lambda windows.](https://proteiniq.io/_next/image?url=%2Fimages%2Fguides%2Ffree-energy-perturbation%2Ffeatured.webp&w=1920&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) Drugs · 18 min ### What is free energy perturbation (FEP) & how it works Free energy perturbation (FEP) predicts how a chemical change alters binding affinity by simulating an alchemical transformation through lambda windows. Learn the thermodynamic cycle, relative vs absolute binding free energy, benchmark accuracy, GPU cost per edge, where FEP fails, and how it compares with docking and MM/GBSA.](https://proteiniq.io/guides/free-energy-perturbation)
- [![Three wireframe simulation windows stacked in an exploded view. The front one shows the backbone trace of ubiquitin inside a shaded periodic water box with a few ions and faint neighbouring periodic images, above a strip of trajectory frames, beside a panel of time-series plots.](https://proteiniq.io/_next/image?url=%2Fimages%2Fguides%2Fmolecular-dynamics-software%2Ffeatured.webp&w=1920&q=75&dpl=dpl_44ysNf1Cbx9iepy8z4SrXP7X4Gnh) Structures · 14 min ### The best free molecular dynamics software in 2026 The best molecular dynamics engines are free: GROMACS, AMBER, OpenMM, NAMD and LAMMPS run most published simulations. We compare them on what each is best at, GPU speed on standard benchmarks, licenses and what a run needs from you, and say where Schrödinger's Desmond still earns its price.](https://proteiniq.io/guides/molecular-dynamics-software)

[See all guides](https://proteiniq.io/guides)

## Frequently asked questions

AI answers from our docs. No sign-in needed.

## Run your first analysis today

Start with 200 free credits and no installs. Upgrade when your project needs more compute.

[Start for free](https://proteiniq.io/sign-up)[View pricing](https://proteiniq.io/pricing)