From f9fbc51c8bdb05298a74720d829a2a58db92c173 Mon Sep 17 00:00:00 2001 From: Shehtab Date: Sat, 2 Aug 2025 21:47:18 -0400 Subject: [PATCH] Patch for processed 3D data loader --- boom/data/load_processed_3d_data.py | 2 +- boom/data/prepare_3D_mols.py | 2 +- boom/datasets/CoordDataset.py | 5 +---- 3 files changed, 3 insertions(+), 6 deletions(-) diff --git a/boom/data/load_processed_3d_data.py b/boom/data/load_processed_3d_data.py index f7765d4..ba71f63 100644 --- a/boom/data/load_processed_3d_data.py +++ b/boom/data/load_processed_3d_data.py @@ -30,6 +30,6 @@ def load_3D_data(property, cached_file, split_file=None): "u298", "zpve", ]: - return retrieve_qm9_dataset(cache_file_name=cached_file) + return retrieve_qm9_dataset(cache_file_name=cached_file, split_file=split_file) else: ValueError("Other 3D datasets currently not tested") diff --git a/boom/data/prepare_3D_mols.py b/boom/data/prepare_3D_mols.py index 4ca2583..9915a9e 100644 --- a/boom/data/prepare_3D_mols.py +++ b/boom/data/prepare_3D_mols.py @@ -77,7 +77,7 @@ def retrieve_qm9_dataset(cache_file_name, split_file): # smiles_data = package_data_loader("qm9_data_with_ood_splits_with_inchi.csv") with open(split_file, "r") as f: - data = f.read().splitlines() + smiles_data = f.read().splitlines() smiles_list = [x.strip().split(",")[0] for x in smiles_data[1:]] smiles_list = [Chem.MolToSmiles(Chem.MolFromSmiles(x)) for x in smiles_list] # inchi_list = [Chem.MolToInchi(Chem.MolFromSmiles(x)) for x in smiles_list] diff --git a/boom/datasets/CoordDataset.py b/boom/datasets/CoordDataset.py index 5985f51..e6a9f7a 100644 --- a/boom/datasets/CoordDataset.py +++ b/boom/datasets/CoordDataset.py @@ -51,13 +51,10 @@ def __init__( split_file = os.path.join( cur_dir, "qm9_data_with_ood_splits_with_inchi.csv" ) - + data = load_processed_data(property, split_file) self.data = load_3D_data(self.property, cached_file, split_file) - self.dataset_type = f"{split.lower()}_{property.lower()}" - data = load_processed_data(property, split_file) - assert type(data) == dict assert type(self.data) == dict