'%20d='M364.9%20143.2v98.9l-1.4%201.2-165.6%20140.4%2035.7%2029.7v-34.2l1.4-1.2%20123.4-105.4%206.5-5.6v37l-1.3%201.2-99.7%2085.8v84.7l-6.6-5.7-97-83.1-3.6-3%203.5-3%20174.4-150.6v-25.1L153.7%20360.5l-6.6%205.7v-98.9l1.4-1.2%20161.6-138-35.7-29.7v34.4L273%20134%20153.7%20236.7l-6.6%205.7v-37l1.3-1.2%2097-83.2V36.3L252%2042l97.1%2083.1%203.6%203-3.6%203-171.7%20147.7V304l106.2-90.6%2056.4-48.8%2018.3-15.7z'%20/%3e%3c/svg%3e)
A guess about physics,written as parameters
Every number your simulation produces is downstream of the force field. It is the one choice you cannot correct later with more sampling.
Bond stretching, angle bending and torsions — cheap to compute and mostly uncontroversial.
Van der Waals and electrostatics. Expensive, long-ranged, and where the disagreements are.
Every atom is assigned a type, and the type is what looks up every parameter above.
Pick bywhat you are simulating
AMBER is not one force field. It is a family, and each member was fitted against a different kind of molecule.
Mixing them within one system is normal and expected — a protein in ff19SB, its ligand in GAFF2, the membrane in lipid21, all in the same box. What is not safe is mixing across families, taking charges derived one way and combining them with van der Waals terms fitted another.
Five steps,and only one is the run
Most of the effort in an AMBER simulation happens before mdrun-equivalent work begins, and most of the mistakes are made there too.
Match the field to the molecule. This decides what the simulation is capable of being right about.
LEaP assigns types and builds topology. Anything non-standard needs charges derived first, with antechamber.
Put the system in a water box, add counter-ions, and decide how much padding you can afford.
Minimise, heat, equilibrate, then produce. The production run is the short description and the long wait.
cpptraj over the trajectory: RMSD, contacts, hydrogen bonds, whatever question you started with.
The parametersare not measurements
A force field is fitted, not derived. Knowing what it was fitted to tells you what it can be trusted about.
Protein force fields are tuned against folded, soluble proteins in water, so they are least reliable exactly where the interesting cases live: disordered regions, unusual protonation states, metal centres, and anything a general small-molecule field like GAFF2 had to guess at. When a result surprises you, the force field is a more likely explanation than the discovery.
AmberMD isbeing built
The image and its definition are being verified. When they are done, it deploys by name like the rest.