ParmFit NCAA

Overview

NCAA is the parmfit(method=ncaa) route for a noncanonical amino acid whose peptide backbone is recognized but whose complete residue graph does not match an active standard Amber protein template. It caps that residue with ACE and NME and builds the capped models with the backbone and the caps frozen, so that only the sidechain R-group relaxes. The α and β conformers are built by rotating the backbone to target (φ,ψ) values of approximately (−60°, −40°) and (−120°, 140°) for an L residue; the signs are reversed for a D residue. ncaa_resids accepts several comma-separated residues; each one is built in its own work directory and all are deployed together. After charge fitting (see ChargeFit), AmberTools prepin/frcmod templates are created, and bond/angle terms plus sidechain torsions are refined. Deployment is validated with tleap — the result block prints the command under Next step, and a clean run reports Errors: 0. Standard residues may also be declared deliberately, to be rebuilt through this flow.

For a worked walkthrough, see the NCAA tutorial.

Target

ncaa_resids must identify protein residues in the input PDB, one selector per comma-separated entry: a chain and residue number such as A123, a residue name and number such as HIF123, or an appended altloc label when a specific alternate conformation is required (for example, A123A for altloc A). A selector without a chain must still be unique. Declare multiple residues by comma-separating their selectors, as in ncaa_resids=HIF89,TRP482. Residue selection occurs after the standard-residue graph match and does not change what the matching recognizes. Every selected residue must have a connected N-H-CA-C(=O) peptide backbone and at least one sidechain heavy atom bonded to CA; ncaa_resnames controls only its output residue name.

The charges in ncaa_charges apply only to the declared residues. Other backbone-only residues in the same PDB receive neither their charge nor a generated NCAA template; the report lists them explicitly for separate handling.

Workflow

The route in outline:

1. Target identity. Locate the target residue and detect backbone, sidechain anchor, and chirality.
2. Capped models. Build and optimize α and β ACE-target-NME conformers.
3. Atomic charges. Fit charges on the optimized conformers with the configured chg_fit route and prepare the charged capped mol2.
4. AmberTools template. Use the fitted charges with antechamber, prepgen, and parmchk2 to create the initial prepin/frcmod files.
5. Parameter refinement. Apply Seminario/mSeminario to bond/angle terms and TorsionFit to target-residue sidechain torsions.
6. Deployment. Write refined prepin/frcmod, processed protein PDB, tleap input, and tleap validation outputs.

Core Parameters

For TorsionFit parameters, QM job resources, MLIP optimization controls, and levels of theory, see the TorsionFit, General Settings, and QM Reference sections of the main ParmFit page.

ChargeFit

NCAA fits charges on the optimized α and β capped conformers. With chg_fit=resp, Gaussian evaluates the ESP of both conformers and AmberTools performs one joint two-stage RESP fit. A charge-capable MLIP or Antechamber method evaluates the two conformers separately, after which MAPLE averages the two per-atom charge sets with equal weights. For model and Antechamber routes, a final backbone adjustment brings the target-residue charge in the prepin to the ncaa_charges value.

chg_fit is independent of iqm. RESP derivation always uses Gaussian for the ESP and AmberTools for the RESP stages; chg_level, chg_route, qm_nproc, and qm_mem control the jobs. NCAA cannot use chg_fit=none because a PDB input does not supply a reusable set of atomic charges.

All charge-route files remain under <base>_work/ncaa/<RESNAME><RESSEQ>/, one directory per declared residue. The resulting charged capped mol2 and target.chg are the inputs to the prepin/GAFF2 template build; the same charges are carried into the Stage 0 parameter set (the bonded-refined starting point) and the TorsionFit baseline. The result block reports Target charge and Actual charge; with several residues it shows only the last one, so check each <rn>_maple.prepin charge sum against its ncaa_charges value.

TorsionFit

NCAA restricts torsion fitting to rotatable dihedrals in the target residue's sidechain; backbone–backbone and cap dihedrals are excluded. torsion_bonds uses residue-local 1-based indices (ACE/NME caps not counted) and separates per-residue groups with ;, in the same order as ncaa_resids. For a complex sidechain, set torsionfit=false and rely on the original GAFF2 torsion, or use torsion_bonds to refine selected dihedrals. NCAA disables torsion ensemble targets internally; see the main ParmFit page for scan and refinement parameter details.

Peptide Boundary Terms

When the NCAA residue is inserted as a custom template, MAPLE dynamically generates ProteinFF-GAFF2 cross-terms at both peptide boundaries: the head side (prev_C–target_N) and the tail side (target_C–next_N). These include exact BOND, ANGLE, and DIHE terms that connect the atom types of the custom template with those of the neighbouring standard residues.

Input Files

Simple Input

A simple NCAA job declares method=ncaa inline and lists its ParmFit options in a sidecar config. The ncaa_resids selectors choose which noncanonical protein residues to build, and the PDB block supplies the complete structure.

#model=uma(size=uma-s-1p1)
#parmfit(method=ncaa,input=NCAA.in)
#device=gpu0

PDB ncaa.pdb

ncaa_resids=A123
ncaa_resnames=MOL
ncaa_charges=0
ncaa_mults=1
chg_fit=resp
bonded=mseminario
torsionfit=false

Full External Input

For a more complex setup, keep the MAPLE launcher short and move ParmFit parameters into an external .parmfit file. Declare method=ncaa inline together with input=; the external config must not set method=. The PDB remains in the MAPLE input; do not place pdb, input, or global MAPLE options in the external config.

#model=uma(size=uma-s-1p1)
#parmfit(method=ncaa,input=NCAA.parmfit)
#device=gpu0

PDB ncaa.pdb

Full ParmFit config

# QM reference
iqm=false
qm_engine=g16
opt_level=B3LYP/def2SVP
sp_level=B3LYP/def2SVP
opt_route=
sp_route=
qm_mode=1
qm_nproc=8
qm_mem=24

# Atomic charges
chg_fit=resp
resp_backend=gaussian
chg_level=HF/6-31G(d)
chg_route=

# Residue declaration, force field, and bonded fitting
ncaa_resids=A123
ncaa_charges=0
ncaa_mults=1
ncaa_resnames=MOL
pro_ff=ff14SB
wat_ff=tip3p
bonded=mseminario
vib_scale=1.0

# Choose the altloc conformer for these residues (selector + letter, e.g. PHE393B)
keep=

# NCAA capped-conformer optimization controls
opt_max_iter=256
opt_max_step=0.2

# TorsionFit
torsionfit=false
torsion_bonds=
torsion_steps=36
backend=cgbs
constraint_mode=projected
stage1_weights=true
torsion_refine_rounds=3
torsion_refine_max_iter=256
torsion_refine_tol=1.0e-6
torsion_ensemble=false
torsion_ensemble_ratio=0.3
torsion_ensemble_weight=0.5

Output Files

• <base>_work/<rn>.prepin and <base>_work/<rn>.frcmod: Initial AmberTools residue template files.
• <base>_work/<rn>_maple.prepin and <base>_work/<rn>_maple.frcmod: Refined MAPLE residue template and parameter files.
• <base>_work/<base>_ncaa_tleap.pdb, <base>_work/<base>_ncaa_tleap.in, and <base>_work/<base>_ncaa_tleap.out: Processed protein PDB, tleap input, and the tleap validation log.
• <base>_work/<base>_ncaa.prmtop, <base>_work/<base>_ncaa.inpcrd, and solvated PDB outputs: tleap validation products.
• <base>_work/ncaa/<RESNAME><RESSEQ>/ — the source residue name and number (HIF89 style, not the exported <rn> name used by the template files above): per-residue optimized conformers, charge-route files, charged capped mol2, target.chg, and AmberTools template intermediates.
• <base>_work/torsionfit/<RESNAME><RESSEQ>/: TorsionFit scans of the declared torsion bonds (scan xyz files and per-bond logs), named with capped-model atom serials.

Requirements and Limits

• The declared residues must have a connected peptide backbone and at least one sidechain heavy atom bonded to CA (see Target); standard atom naming is not required.
• chg_fit=none is unsupported because NCAA has no input mol2 charges to preserve.
• When chg_fit=resp, RESP uses the Gaussian backend; model and Antechamber charge routes do not require RESP.
• The final target-residue charge written for model and Antechamber methods is constrained to the ncaa_charges value.
• TorsionFit is restricted to rotatable dihedrals in the target residue's sidechain and does not fit peptide backbone torsions.