ValueError: No template found for N-terminal residue in large multi-chain system (150 chains)

[Parwez4436]

I am attempting to set up a large-scale Coarse-Grained simulation using the Open-AWSEM forcefield in OpenMM. The system consists of 150 copies of the Mouse Prion Protein (residues 23-225, ~203 residues per chain) arranged in a slab geometry (25×25×150 nm).

I am encountering two primary bottlenecks when using PDBFixer and OpenMM's ForceField.createSystem():

1. ValueError: No template found for residue 1 (LYS).
   Even after using PDBFixer to standardize the topology, OpenMM fails to match the N-terminal Lysine. The error persists even if Hydrogens are removed. It seems the internal _matchAllResiduesToTemplates logic is struggling with the terminal connectivity in this multi-chain setup.

2. OOM / Process Killed during addMissingAtoms()
   When running fixer.addMissingAtoms() on the full 150-chain system (~170,000 atoms/beads), the process is killed by the OS (Out of Memory). It appears the LocalEnergyMinimizer or the internal system creation within PDBFixer does not scale linearly with the number of chains/residues in terms of RAM usage.

System Details:

• OpenMM Version: 8.x (via Miniconda)
• PDBFixer Version: 1.9
• System Size: 150 chains, ~203 residues/chain.
• Input: Custom PDB generated by replicating a single-chain template.

Error Traceback:

ValueError: No template found for residue 1 (LYS). This might mean your input topology is missing some atoms or bonds, or possibly that you are using the wrong force field. 
For more information, see https://github.com/openmm/openmm/wiki/Frequently-Asked-Questions#template

What I have tried:

• Standardizing a single chain with PDBFixer first, then replicating.
• Ensuring TER records are present after every chain.
• Using chains=None in OpenMMAWSEMSystem (Open-AWSEM wrapper for app.PDBFile).
• Strict column alignment in the PDB file.

Questions for the Developers:

1. Is there a way to force PDBFixer to skip the expensive minimization step during addMissingAtoms() to prevent OOM for large systems?
2. In multi-chain systems where Chain IDs must repeat (exceeding the 62-character A-Z, a-z, 0-9 limit), does the template matcher lose track of terminal residues (N-TER/C-TER)?
3. Are there specific atom-naming requirements for the AWSEM 6-bead model (N, H, CA, C, O, CB) to be recognized as a valid N-terminal template without causing a mismatch?

[epretti]

Can you attach your PDB file(s) and any other required input files along with a self-contained script that reproduces the problem?

[peastman]

I'm not familiar with OpenMMAWSEMSystem, but I suspect the Topology is inconsistent with the force field in some way. Take a look at the FAQ entry linked in the error message, and also the section of the user guide that describes how force fields match templates to residues. PDBFixer is going to create an all atom model, not a coarse grained one. You'll need to process the resulting Topology to convert it to the coarse grained representation you want to use, and make sure it's exactly consistent with what's described in the force field.

Is there a way to force PDBFixer to skip the expensive minimization step during addMissingAtoms() to prevent OOM for large systems?

The minimization is an essential part. Without it, you'll get atoms in nonsensical positions.

In multi-chain systems where Chain IDs must repeat (exceeding the 62-character A-Z, a-z, 0-9 limit), does the template matcher lose track of terminal residues (N-TER/C-TER)?

As long as you have TER records, it will recognize the chain breaks. If you're ever in doubt, inspect the Topology object to see exactly what it describes.

Are there specific atom-naming requirements for the AWSEM 6-bead model (N, H, CA, C, O, CB) to be recognized as a valid N-terminal template without causing a mismatch?

See the FAQ entry. It discusses where naming does and doesn't matter.

[Parwez4436]

@peastman @epretti, thank you for your patience.

I have attached the requested files to reproduce the ValueError: No template found for residue 0 (LYS):

1. PrP_2chain_debug.pdb: A minimal 2-chain version of my system.
2. openawsem.xml: The forcefield file used in my environment.
3. reproduce_issue.py: A self-contained script to trigger the error.

In the PDB, the LYS residue contains atoms [N, CA, C, O, CB]. This is a Coarse-Grained model where Hydrogens are intentionally absent. The error persists even when using ignoreExternalBonds=True.

I am looking for a way to make the OpenMM template matcher accept this topology without reporting missing atoms or bond inconsistencies. Thank you for your help!

reproduce_issue.py
openawsem.xml
PrP_2chain_debug.pdb

[peastman]

Your residue templates don't include any bonds. They just list the atoms, but not how they are connected to each other. Since the Topology you're trying to match to does include bonds (PDB automatically adds bonds within standard amino acids), none of them can match.

See the discussion in the manual, which describes how templates are matched to residues.