Tutorials¶
Complete runs, start to finish, each one executed exactly as written. The output shown on every page is copied from the files that run produced, not composed for the page.
Every page states the md-tools version it was executed against, at the top. A page whose version is older than the one you have installed has not been re-run for your release: the commands may still be right, but nothing on this site claims they were checked against it.
At each major release every tutorial is re-run as written and adjusted where it no longer passes. The 0.5.3 pages, and the 0.6.4 paracetamol clustering page that the symmetry-first workflow replaced, are archived, with the reasons.
Start from a system¶
Each system page introduces the molecule, shows its structure, and builds it — explicit or implicit solvent — before handing you to a method. The systems are ordered by what they are for, not by size.
| system | what it is for | methods |
|---|---|---|
| Alanine dipeptide | the toy: two torsions with barriers a plain run crosses, so a method can be checked against the truth | cMD, REST2, AIS |
| Paracetamol | a ligand on its own: parameterise once, reuse everywhere | cMD, REST2, AIS |
| Ligand parameterization | how a small molecule is prepared before any run: protonation at pH 7, parameters, and which torsions REST2 heats — paracetamol as a ligand, c(RGDfV) as a peptide-like molecule | parameterization, REST2 scaler |
| c(RGDfV) | a cyclic peptide: its conformations from the unscaled state of a REST2 ladder, with two symmetry operations on the torsions | analysis of registered REST2 data |
| Chignolin | a folding peptide: a real equilibrium, small enough to compare methods in a day | cMD, REST2 |
| TYK2 + ejm_31 | a kinase with a real inhibitor: where a hot region is worth CHOOSING | cMD, selective REST2, AIS |
The methods, and where they are¶
| method | systems | state |
|---|---|---|
| cMD | alanine dipeptide, paracetamol, chignolin, TYK2 (1 µs) | published |
| REST2 | alanine dipeptide, paracetamol, chignolin, TYK2 (selective) | published |
| AIS | alanine dipeptide, paracetamol; TYK2 as a measured LIMIT | published |
| Umbrella sampling | alanine dipeptide (φ), protein–ligand (centre-of-mass distance) | page planned |
A planned row is listed so the shape of the set is visible; it is never linked to a page that does not exist. The tutorials are about conformational sampling; no alchemical transformation and no protein–protein complex is taught here.
Run on NVIDIA RTX 3080 GPUs with CUDA and mixed precision.
| page | version it was run against |
|---|---|
| ligand parameterization | 0.6.5 development branch (171a708e), CPU only |
| c(RGDfV) / analysis | 0.6.5 plus the unreleased alignment methods, on a dataset run with 0.5.4 |
| paracetamol / torsions in one labelling | 0.6.5 plus the unreleased alignment methods, on the 0.6.1 runs |
| paracetamol / cMD | 0.6.1 |
| paracetamol / REST2 | 0.6.1 |
| paracetamol / AIS | 0.6.1 |
| chignolin / cMD | 0.6.1 |
| chignolin / REST2 | 0.6.1 |
| alanine dipeptide / AIS | 0.6.1 |
| TYK2 + ejm_31 / selective REST2 | 0.6.1 |
| alanine dipeptide / cMD | 0.6.1 |
| alanine dipeptide / REST2 | 0.6.1 |
| TYK2 + ejm_31 / AIS | 0.6.1 |
| TYK2 + ejm_31 / cMD | 0.6.1 |
| chignolin / choosing τ_max | 0.5.4, not re-run |