Hi Enrico,

  This error means that the Boltz neural net weights for affinity prediction

        ~/.boltz/boltz2_aff.ckpt

either failed to be completely downloaded when you installed Boltz.  This can happen if you quit ChimeraX during the Boltz installation before it completed.  The easiest way to fix the problem is to delete the ~/.boltz directory.  Then the next time you run a prediction it will download the weights for structure prediction and affinity prediction and the chemical component dictionary.  Those files are about 6 Gbytes so it may take a while to download (minutes to hours depending on your network speed).

        Tom

> On Sep 2, 2025, at 2:02 AM, Enrico Martinez via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
>
> Dear ChimeraX users!
>
> I am testing the Boltz-2 on Mac in the ChimeraX-daily: for one receptor (provided fasta sequence) and 5 ligands (provided smiles strings for each compound). My calculations are running on GPUs and I see the Boltz-2 folder with the results. However, the ChimeraX send the following error message (copied directly from log):
>
> Running Boltz prediction of 5 ligands on gpu
> Using cached multiple sequence alignment
> Running boltz prediction failed with exit code 1:
> command:
> /Users/enrico/boltz22/bin/boltz predict /Users/enrico/Desktop/boltz_2 --accelerator gpu --no_kernels
> stdout:
> Boltz version 2.2.0
> Checking input data.
> Processing 5 inputs with 5 threads.
> Running structure prediction for 5 inputs.
> Predicting: | | 0/? [00:00<?, ?it/s] Predicting: 0%| | 0/5 [00:00<?, ?it/s] Predicting DataLoader 0: 0%| | 0/5 [00:00<?, ?it/s] Predicting DataLoader 0: 20%|██ | 1/5 [01:55<07:42, 0.01it/s] Predicting DataLoader 0: 40%|████ | 2/5 [03:55<05:53, 0.01it/s] Predicting DataLoader 0: 60%|██████ | 3/5 [05:53<03:55, 0.01it/s] Predicting DataLoader 0: 80%|████████ | 4/5 [07:56<01:59, 0.01it/s] Predicting DataLoader 0: 100%|██████████| 5/5 [09:59<00:00, 0.01it/s]Number of failed examples: 0
> Predicting DataLoader 0: 100%|██████████| 5/5 [09:59<00:00, 0.01it/s]
>
> Predicting property: affinity
>
> Checking input data for affinity.
> Running affinity prediction for 5 inputs.
>
> stderr:
> 0%| | 0/5 [00:00<?, ?it/s] 20%|██ | 1/5 [00:05<00:23, 5.78s/it] 100%|██████████| 5/5 [00:05<00:00, 1.17s/it]
> GPU available: True (mps), used: True
> TPU available: False, using: 0 TPU cores
> HPU available: False, using: 0 HPUs
> /Users/enrico/boltz22/lib/python3.11/site-packages/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py:76: Starting from v1.9.0, `tensorboardX` has been removed as a dependency of the `pytorch_lightning` package, due to potential conflicts with other packages in the ML ecosystem. For this reason, `logger=True` will use `CSVLogger` as the default logger, unless the `tensorboard` or `tensorboardX` packages are found. Please `pip install lightning[extra]` or one of them to enable TensorBoard support by default
> Mon Sep 1 14:38:24 2025: Loading Boltz structure prediction weights
> /Users/enrico/boltz22/lib/python3.11/site-packages/pytorch_lightning/utilities/migration/utils.py:56: The loaded checkpoint was produced with Lightning v2.5.0.post0, which is newer than your current Lightning version: v2.5.0
> Mon Sep 1 14:38:35 2025: Finished loading Boltz structure prediction weights
> Mon Sep 1 14:38:35 2025: Starting structure inference
> /Users/enrico/boltz22/lib/python3.11/site-packages/pytorch_lightning/trainer/connectors/data_connector.py:420: Consider setting `persistent_workers=True` in 'predict_dataloader' to speed up the dataloader worker initialization.
> /Users/enrico/boltz22/lib/python3.11/site-packages/torch/utils/data/dataloader.py:684: UserWarning: 'pin_memory' argument is set as true but not supported on MPS now, then device pinned memory won't be used.
> warnings.warn(warn_msg)
> Mon Sep 1 14:38:44 2025: Begin structure inference ligand4 (1 of 5)
> /Users/enrico/boltz22/lib/python3.11/site-packages/boltz/model/loss/diffusionv2.py:51: UserWarning: The operator 'aten::linalg_svd' is not currently supported on the MPS backend and will fall back to run on the CPU. This may have performance implications. (Triggered internally at /Users/runner/work/pytorch/pytorch/pytorch/aten/src/ATen/mps/MPSFallback.mm:15.)
> U, S, V = torch.linalg.svd(
> Mon Sep 1 14:40:40 2025: End structure inference ligand4 (1 of 5)
> Mon Sep 1 14:40:40 2025: Begin structure inference ligand5 (2 of 5)
> Mon Sep 1 14:42:40 2025: End structure inference ligand5 (2 of 5)
> Mon Sep 1 14:42:40 2025: Begin structure inference ligand2 (3 of 5)
> Mon Sep 1 14:44:38 2025: End structure inference ligand2 (3 of 5)
> Mon Sep 1 14:44:38 2025: Begin structure inference ligand3 (4 of 5)
> Mon Sep 1 14:46:41 2025: End structure inference ligand3 (4 of 5)
> Mon Sep 1 14:46:41 2025: Begin structure inference ligand1 (5 of 5)
> Mon Sep 1 14:48:44 2025: End structure inference ligand1 (5 of 5)
> Mon Sep 1 14:48:57 2025: Finished structure inference
> Mon Sep 1 14:48:57 2025: Loading affinity prediction weights
> Traceback (most recent call last):
> File "/Users/enrico/boltz22/bin/boltz", line 8, in <module>
> sys.exit(cli())
> ^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/click/core.py", line 1157, in __call__
> return self.main(*args, **kwargs)
> ^^^^^^^^^^^^^^^^^^^^^^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/click/core.py", line 1078, in main
> rv = self.invoke(ctx)
> ^^^^^^^^^^^^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/click/core.py", line 1688, in invoke
> return _process_result(sub_ctx.command.invoke(sub_ctx))
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/click/core.py", line 1434, in invoke
> return ctx.invoke(self.callback, **ctx.params)
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/click/core.py", line 783, in invoke
> return __callback(*args, **kwargs)
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/boltz/main.py", line 1419, in predict
> model_module = Boltz2.load_from_checkpoint(
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/pytorch_lightning/utilities/model_helpers.py", line 125, in wrapper
> return self.method(cls, *args, **kwargs)
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/pytorch_lightning/core/module.py", line 1581, in load_from_checkpoint
> loaded = _load_from_checkpoint(
> ^^^^^^^^^^^^^^^^^^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/pytorch_lightning/core/saving.py", line 63, in _load_from_checkpoint
> checkpoint = pl_load(checkpoint_path, map_location=map_location)
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/lightning_fabric/utilities/cloud_io.py", line 60, in _load
> return torch.load(
> ^^^^^^^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/torch/serialization.py", line 1491, in load
> with _open_zipfile_reader(opened_file) as opened_zipfile:
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> File "/Users/enrico/boltz22/lib/python3.11/site-packages/torch/serialization.py", line 771, in __init__
> super().__init__(torch._C.PyTorchFileReader(name_or_buffer))
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> RuntimeError: PytorchStreamReader failed reading zip archive: failed finding central directory
>
> Could you precise which package causes this issue and how would the problem be solved in the ChimeraX update?
>
> Many thanks for help!!
>
> Enrico
> _______________________________________________
> ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu
> To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu
> Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/