AttributeError: "'seqemb_mode_enabled'"

[QUEST2179]

The error seems from openfold-2.0.0, but how to fix it without downgrading openfold? Thanks!

python projects/tfold_ag/predict.py --pid_fpath=examples/prot_ids.nanoag.txt --fas_dpath=examples/fasta.files --msa_fpath=examples/msa.files/7sai_A.a3m --pdb_dpath=examples/pdb.files.nano
[09/28 07:01:04 root]: restoring the pre-trained tFold-PPI-SeqPT model ...
[09/28 07:01:08 root]: model weights restored from params/esm_ppi_650m.pth
[09/28 07:01:08 root]: restoring the pre-trained AlphaFold2 model ...
Traceback (most recent call last):
File "/home/miniconda/envs/gpu9/lib/python3.9/site-packages/ml_collections/config_dict/config_dict.py", line 903, in getitem
field = self._fields[key]
KeyError: 'seqemb_mode_enabled'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/home/miniconda/envs/gpu9/lib/python3.9/site-packages/ml_collections/config_dict/config_dict.py", line 827, in getattr
return self[attribute]
File "/home/miniconda/envs/gpu9/lib/python3.9/site-packages/ml_collections/config_dict/config_dict.py", line 909, in getitem
raise KeyError(self._generate_did_you_mean_message(key, str(e)))
KeyError: "'seqemb_mode_enabled'"

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/home/antibody/tfold-master/projects/tfold_ag/predict.py", line 81, in
main()
File "/home/antibody/tfold-master/projects/tfold_ag/predict.py", line 73, in main
predict(args.pid_fpath, args.fas_dpath,
File "/home/antibody/tfold-master/projects/tfold_ag/predict.py", line 62, in predict
predictor = AgPredictor(
File "/home/antibody/tfold-master/./tfold/deploy/ag_predictor.py", line 32, in init
self.psp_featurizer = PspFeaturizer.restore(psp_path)
File "/home/antibody/tfold-master/./tfold/deploy/psp_featurizer.py", line 51, in restore
model = cls(config).eval()
File "/home/antibody/tfold-master/./tfold/deploy/psp_featurizer.py", line 39, in init
self.model = AlphaFold(self.config)
File "/home/miniconda/envs/gpu9/lib/python3.9/site-packages/openfold-2.0.0-py3.9-linux-x86_64.egg/openfold/model/model.py", line 84, in init
self.seqemb_mode = config.globals.seqemb_mode_enabled
File "/home/miniconda/envs/gpu9/lib/python3.9/site-packages/ml_collections/config_dict/config_dict.py", line 829, in getattr
raise AttributeError(e)
AttributeError: "'seqemb_mode_enabled'"

[Wolkenwandler]

The error seems from openfold-2.0.0, but how to fix it without downgrading openfold? Thanks!

python projects/tfold_ag/predict.py --pid_fpath=examples/prot_ids.nanoag.txt --fas_dpath=examples/fasta.files --msa_fpath=examples/msa.files/7sai_A.a3m --pdb_dpath=examples/pdb.files.nano [09/28 07:01:04 root]: restoring the pre-trained tFold-PPI-SeqPT model ... [09/28 07:01:08 root]: model weights restored from params/esm_ppi_650m.pth [09/28 07:01:08 root]: restoring the pre-trained AlphaFold2 model ... Traceback (most recent call last): File "/home/miniconda/envs/gpu9/lib/python3.9/site-packages/ml_collections/config_dict/config_dict.py", line 903, in getitem field = self._fields[key] KeyError: 'seqemb_mode_enabled'

During handling of the above exception, another exception occurred:

Traceback (most recent call last): File "/home/miniconda/envs/gpu9/lib/python3.9/site-packages/ml_collections/config_dict/config_dict.py", line 827, in getattr return self[attribute] File "/home/miniconda/envs/gpu9/lib/python3.9/site-packages/ml_collections/config_dict/config_dict.py", line 909, in getitem raise KeyError(self._generate_did_you_mean_message(key, str(e))) KeyError: "'seqemb_mode_enabled'"

During handling of the above exception, another exception occurred:

Traceback (most recent call last): File "/home/antibody/tfold-master/projects/tfold_ag/predict.py", line 81, in main() File "/home/antibody/tfold-master/projects/tfold_ag/predict.py", line 73, in main predict(args.pid_fpath, args.fas_dpath, File "/home/antibody/tfold-master/projects/tfold_ag/predict.py", line 62, in predict predictor = AgPredictor( File "/home/antibody/tfold-master/./tfold/deploy/ag_predictor.py", line 32, in init self.psp_featurizer = PspFeaturizer.restore(psp_path) File "/home/antibody/tfold-master/./tfold/deploy/psp_featurizer.py", line 51, in restore model = cls(config).eval() File "/home/antibody/tfold-master/./tfold/deploy/psp_featurizer.py", line 39, in init self.model = AlphaFold(self.config) File "/home/miniconda/envs/gpu9/lib/python3.9/site-packages/openfold-2.0.0-py3.9-linux-x86_64.egg/openfold/model/model.py", line 84, in init self.seqemb_mode = config.globals.seqemb_mode_enabled File "/home/miniconda/envs/gpu9/lib/python3.9/site-packages/ml_collections/config_dict/config_dict.py", line 829, in getattr raise AttributeError(e) AttributeError: "'seqemb_mode_enabled'"

Hello, thank you for your attention to the work of tfold. You can download the latest code for running. The new version of the code has removed the dependency on the openfold package. Using the new code should not encounter the above problems.