mmCIF file missing required field _entity_poly_seq.mon_id

Hi there,

We are attempting to configure ColabFold to run on an HPC system without internet access using a Singularity container. This requires that we make no external queries and work only with local data. The container is deployed as follows:

singularity run -B $COLABFOLD_DATA_PATH:/cache/colabfold --nv $COLABFOLD_CONTAINER \
        colabfold_batch \
        --data /path/to/alphafold2/params/ \
        --templates \
        --custom-template-path "/path/to/alphafold2/mmcif_files" \
        --local-pdb-path "/cache/colabfold/pdb/divided" \
        --num-models 1 \
        --model-type "alphafold2_multimer_v3" \
        --amber \
        --use-gpu-relax \
        --random-seed 0 \
        <input> <results>

After some time, the run fails with the following:

ValueError: mmCIF file /path/to/alphafold2/mmcif_files/1c58.cif is missing required field _entity_poly_seq.mon_id

We have sourced the alphafold2 mmCIF files using the methods detailed in the public AF2 repo: https://github.com/google-deepmind/alphafold/blob/main/scripts/download_pdb_mmcif.sh as well as directly from https://colabfold.mmseqs.com/. Is there a different accessible source for this data with the required fields?

Start with the failing file 1c58.cif and the AlphaFold download script and ColabFold source links in the report. Check whether the locally supplied mmCIF contains _entity_poly_seq.mon_id and compare the two reported data sources. Done means identifying a usable source with the required field or documenting the incompatibility clearly.