Reopen #7: pLDDT not writing to PDB

[LandonGetz]

I am having the same issue as was previously reported in #7, where all pLDDT values are being stored as 1.00. Including print(protein.plddt) provides the per-residue tensor values for pLDDT (values less than 1), so I know the model is generating them. I am running code that is very similar to #7.

Am I missing something obvious or has that previous bug popped back up?

[LandonGetz]

I think this issue is in handling of the protein_complex into .to_pdb. If I print the output of

protein_complex= protein.to_protein_complex()
for chain in protein_complex.chain_iter():
    print(chain.confidence)

I get an array of 1's (which is what is being output into the b-factor column of my PDBs). I've managed to hack my way out of this by running protein.to_pdb_string() and then writing the PDB file manually:

scaled_plddt = protein.plddt * 100
 protein_chain = protein.to_protein_chain()
 protein_chain.confidence = scaled_plddt.detach().cpu().numpy()
 pdb_string = protein_chain.to_pdb_string()
 with open(out_pdb, "w") as f:
        f.write(pdb_string)

Let me know if I'm just missing something silly, or if this is a bona fide bug

[anjalijain22]

Hello, thank you so much for developing and maintaining this tool! I am facing the same issue here where the pLDDT scores are not being written to the PDB file. Currently, @LandonGetz's hack works for me (thank you very much!) but am wondering if there is a fix in the works for this?

[ebetica]

cc @imathur1 , I think this is a real bug?

[imathur1]

Similar to this issue https://github.com/evolutionaryscale/esm/issues/247, I believe this PR https://github.com/evolutionaryscale/esm/pull/276/files#diff-253d59232c58fceb8e86ac22dbcd860107afb9c3d5c4518d9b08ce997a9aab7e addresses the problem of all 1s being written out to the PDB file in the b factor column.

I ran this snippet

    model: ESM3InferenceClient = ESM3.from_pretrained("esm3-sm-open-v1").to("cuda")
    prompt = "HERPYACP_________________________________________________________________________________"
    protein = ESMProtein(sequence=prompt)
    protein = model.generate(protein, GenerationConfig(track="sequence", num_steps=8, temperature=0.5))
    protein = model.generate(protein, GenerationConfig(track="structure", num_steps=8))
    protein.to_pdb("protein.pdb") # type: ignore

from https://github.com/evolutionaryscale/esm/issues/7 and I see the PLDDT values are not 1 in the PDB file:

ATOM      1  N   HIS A   1     -25.586   9.408  10.481  1.00  0.89           N  
ATOM      2  CA  HIS A   1     -24.918   8.158  10.135  1.00  0.89           C  
ATOM      3  C   HIS A   1     -24.927   7.187  11.311  1.00  0.89           C  
ATOM      4  N   GLU A   2     -25.360   6.203  11.210  1.00  0.94           N  
ATOM      5  CA  GLU A   2     -25.428   5.181  12.249  1.00  0.94           C  
ATOM      6  C   GLU A   2     -24.157   4.338  12.274  1.00  0.94           C  
ATOM      7  N   ARG A   3     -23.591   3.877  13.252  1.00  0.96           N  
ATOM      8  CA  ARG A   3     -22.415   3.034  13.439  1.00  0.96           C  
ATOM      9  C   ARG A   3     -22.749   1.803  14.275  1.00  0.96           C  
ATOM     10  N   PRO A   4     -23.260   0.926  13.716  1.00  0.96           N  
ATOM     11  CA  PRO A   4     -23.783  -0.184  14.505  1.00  0.96           C  
ATOM     12  C   PRO A   4     -22.705  -1.235  14.755  1.00  0.96           C  
ATOM     13  N   TYR A   5     -21.536  -1.133  14.074  1.00  0.97           N  
ATOM     14  CA  TYR A   5     -20.496  -2.128  14.313  1.00  0.97           C  
ATOM     15  C   TYR A   5     -19.547  -1.677  15.417  1.00  0.97           C  
ATOM     16  N   ALA A   6     -19.575  -2.323  16.450  1.00  0.98           N  
ATOM     17  CA  ALA A   6     -18.739  -1.976  17.595  1.00  0.98           C  

Let me know if an issue still persists!