I am testing DiffDock on a set of ~9K small molecules from DUD-e.
I managed to cache the ESM embedding, but still every iteration takes around 60 seconds to complete.
I notinced in the output of the code multiple WARNING errors of this kind:
17it [16:32, 60.05s/it]
[2025-May-20 08:31:16 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 1: 1 / 10 samples failed
[2025-May-20 08:31:19 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 2: 1 / 10 samples failed
[2025-May-20 08:31:23 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 3: 1 / 10 samples failed
[2025-May-20 08:31:27 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 4: 1 / 10 samples failed
[2025-May-20 08:31:30 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 5: 1 / 10 samples failed
[2025-May-20 08:31:33 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 6: 1 / 10 samples failed
[2025-May-20 08:31:36 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 7: 1 / 10 samples failed
[2025-May-20 08:31:39 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 8: 1 / 10 samples failed
[2025-May-20 08:31:42 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 9: 1 / 10 samples failed
[2025-May-20 08:31:44 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 10: 1 / 10 samples failed
[2025-May-20 08:31:46 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 11: 1 / 10 samples failed
[2025-May-20 08:31:49 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 12: 1 / 10 samples failed
[2025-May-20 08:31:51 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 13: 1 / 10 samples failed
[2025-May-20 08:31:53 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 14: 1 / 10 samples failed
[2025-May-20 08:31:54 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 15: 1 / 10 samples failed
[2025-May-20 08:31:56 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 16: 1 / 10 samples failed
[2025-May-20 08:31:58 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 17: 1 / 10 samples failed
[2025-May-20 08:32:00 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 18: 1 / 10 samples failed
19it [18:24, 58.15s/it]
I could not find where the Warning is raised despite diving in the code for a lot of time. In the end the protein-ligand complexes are still calculated, but this is slowing down the process.
Do someone know how to fix this? any suggestion appreciated
I am testing DiffDock on a set of ~9K small molecules from DUD-e.
I managed to cache the ESM embedding, but still every iteration takes around 60 seconds to complete.
I notinced in the output of the code multiple WARNING errors of this kind:
17it [16:32, 60.05s/it]
[2025-May-20 08:31:16 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 1: 1 / 10 samples failed
[2025-May-20 08:31:19 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 2: 1 / 10 samples failed
[2025-May-20 08:31:23 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 3: 1 / 10 samples failed
[2025-May-20 08:31:27 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 4: 1 / 10 samples failed
[2025-May-20 08:31:30 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 5: 1 / 10 samples failed
[2025-May-20 08:31:33 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 6: 1 / 10 samples failed
[2025-May-20 08:31:36 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 7: 1 / 10 samples failed
[2025-May-20 08:31:39 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 8: 1 / 10 samples failed
[2025-May-20 08:31:42 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 9: 1 / 10 samples failed
[2025-May-20 08:31:44 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 10: 1 / 10 samples failed
[2025-May-20 08:31:46 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 11: 1 / 10 samples failed
[2025-May-20 08:31:49 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 12: 1 / 10 samples failed
[2025-May-20 08:31:51 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 13: 1 / 10 samples failed
[2025-May-20 08:31:53 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 14: 1 / 10 samples failed
[2025-May-20 08:31:54 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 15: 1 / 10 samples failed
[2025-May-20 08:31:56 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 16: 1 / 10 samples failed
[2025-May-20 08:31:58 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 17: 1 / 10 samples failed
[2025-May-20 08:32:00 UTC] WARNING -Complex ['ALK5_1_ZINC20760439'] Batch 1 Inference Iteration 18: 1 / 10 samples failed
19it [18:24, 58.15s/it]
I could not find where the Warning is raised despite diving in the code for a lot of time. In the end the protein-ligand complexes are still calculated, but this is slowing down the process.
Do someone know how to fix this? any suggestion appreciated