Skip to content

Ethane → chloroethane

On hold

Alchemical transformation is on hold as of 2026-10-13 and is not part of the current release. This page was last run against 0.6.4. See On hold.

Tested against md-tools 0.6.4. Every command and every number on these pages comes from a run executed as written, at that version, on NVIDIA RTX 3080 GPUs with CUDA and mixed precision.

Two molecules that differ by one atom, and the smallest relative free energy calculation there is. This is the pattern every protein–ligand campaign uses: rather than computing two absolute free energies and subtracting, mutate one molecule into the other and measure only the difference.

Why this pair

  • The difference is the whole point. Relative free energy is cheaper and better-conditioned than two absolutes, because everything the two molecules have in common cancels instead of being sampled twice. Seven of ethane's eight atoms map onto chloroethane; only the mutating end is alchemical.
  • Both are in FreeSolv, so the answer can be checked against experiment: ethane +1.83, chloroethane −0.63 kcal/mol, a difference of −2.46.
  • There is a second, independent route to the same number. Decoupling each molecule from water separately gives two absolute hydration free energies whose difference is the same quantity — computed with a different plan, a different Hamiltonian and a different number of legs. That cross-check is in the alchemical validation page, and it is worth more than agreement with experiment.
  • It is small. Six ladders — two legs, three repeats each — finish in about two hours across four cards.

Two molecules, two boxes, two legs

This is the structural difference from absolute hydration, and everything else follows from it:

   ethane(vac) --[ dG_vac ]--> chloroethane(vac)
        |                            |
   dG_hyd(ethane)            dG_hyd(chloroethane)
        v                            v
   ethane(sol) --[ dG_sol ]--> chloroethane(sol)

   ddG_hyd = dG_hyd(chloroethane) - dG_hyd(ethane) = dG_sol - dG_vac

Neither leg is a physical process and neither ΔG is measurable. Only the difference is, and only because the cycle closes. A mutation in vacuum has no meaning on its own — but subtract it from the same mutation in water and what remains is the difference of two hydration free energies.

The vacuum leg is NOT zero here. In the decoupling tutorial the molecule's own Hamiltonian is retained unchanged at both ends, so the vacuum leg is λ-independent by construction and is not run. Here the molecule itself changes, so its internal energy changes with λ and that leg has to be sampled.

So you build two boxes — ethane in vacuum and ethane in water — and run two legs of one cycle. Both legs live under ONE <system>, and the layout says which is which:

<system>/
  input/RHFE-leg1.in          the solvent leg's input, shared by its repeats
  input/RHFE-leg2.in          the vacuum leg's
  RHFE-leg1-run1/  RHFE-leg1-run2/  RHFE-leg1-run3/
  RHFE-leg2-run1/  RHFE-leg2-run2/  RHFE-leg2-run3/

alchemical.cycle and alchemical.leg name the run, exactly as every other protocol's runs are <method>-run<N>. The cycle is declared, not derived, and the reason is worth knowing: an RBFE's solvent leg and an RHFE's solvent leg are the same calculation on the same box. What distinguishes them is the OTHER leg. Deriving the cycle from one leg would be guessing from evidence that does not contain the answer.

Build them

Two build-top runs from the same parameter package, differing only in the solvent block. Both belong to the same <system>; they produce the two boxes the two legs run in:

# vacuum/build.config
solute:
  kind: ligand
  residue_name: ETA
  parameters: ./parameter        # a PATH, relative to this file
solvent:
  model: vacuum
constraints:
  type: HBonds

# solvent/build.config
solute:
  kind: ligand
  residue_name: ETA
  parameters: ./parameter
solvent:
  model: TIP3P
  padding_nm: 1.35
  box_shape: cube
  ionic_strength_molar: 0.0
  cutoff_nm: 0.9
constraints:
  type: HBonds
  rigid_water: true

Both boxes hold ethane — endpoint A. Chloroethane never has a box of its own: it is appended into ethane's, through the atom map, when the plan is built. Two independently solvated boxes would not be atom-matched, and combine-topology refuses that rather than trying to reconcile them.

Where to go next

method what it is for
Relative hydration free energy ΔΔG_hyd by mutating one molecule into the other published
Relative binding free energy the same machinery in a binding site planned, 0.6.5

A row with no link is a page that does not exist yet.