LBA / lba_stats.json
Ty-Perez's picture
Upload lba_stats.json with huggingface_hub
b369c0c verified
Raw
History Blame Contribute Delete
13.4 kB
{
"n_total": 4463,
"sample_size": 4463,
"atoms_per_structure": {
"min": 115,
"max": 1085,
"mean": 397.0734931660318
},
"bounding_box_extent_angstrom": {
"side_min": 10.526999950408936,
"side_max": 47.270999908447266,
"diagonal_min": 25.27603114136397,
"diagonal_max": 59.69870791297603
},
"elements": {
"count": 23,
"atomic_numbers": [
1,
6,
7,
8,
9,
11,
12,
15,
16,
17,
19,
20,
25,
26,
27,
28,
29,
30,
35,
38,
48,
53,
55
],
"symbols": [
"H",
"C",
"N",
"O",
"F",
"Na",
"Mg",
"P",
"S",
"Cl",
"K",
"Ca",
"Mn",
"Fe",
"Co",
"Ni",
"Cu",
"Zn",
"Br",
"Sr",
"Cd",
"I",
"Cs"
],
"counts": {
"1": 682255,
"6": 702697,
"7": 178352,
"8": 197780,
"9": 995,
"11": 26,
"12": 177,
"15": 824,
"16": 7325,
"17": 659,
"19": 3,
"20": 61,
"25": 59,
"26": 27,
"27": 26,
"28": 8,
"29": 4,
"30": 686,
"35": 130,
"38": 1,
"48": 1,
"53": 42,
"55": 1
}
},
"label_y": {
"min": 2.0,
"max": 11.920000076293945,
"mean": 6.453963702605921
},
"exact": {
"num_complexes": 4463,
"unique_pdb_ids": 4463,
"components": {
"protein": {
"total_atoms": 27769649,
"min": 566,
"mean": 6222.1934,
"max": 74903,
"zero_length": 0
},
"pocket": {
"total_atoms": 1658646,
"min": 100,
"mean": 371.6437,
"max": 1021,
"zero_length": 0
},
"ligand": {
"total_atoms": 113493,
"min": 6,
"mean": 25.4298,
"max": 71,
"zero_length": 0
}
},
"pocket_provenance": "ATOM3D's OWN published selection (protein atoms within 6.0 A of any ligand atom); STORED because it is the source convention the seq-id benchmarks are defined on, not a crop we invented. It is a RESIDUE shell (0 of 122,736 pocket residues truncated) and it EXCLUDES water entirely.",
"elements_z": [
1,
3,
6,
7,
8,
9,
11,
12,
15,
16,
17,
19,
20,
25,
26,
27,
28,
29,
30,
34,
35,
38,
48,
53,
55,
80
],
"elements_symbols": [
"H",
"Li",
"C",
"N",
"O",
"F",
"Na",
"Mg",
"P",
"S",
"Cl",
"K",
"Ca",
"Mn",
"Fe",
"Co",
"Ni",
"Cu",
"Zn",
"Se",
"Br",
"Sr",
"Cd",
"I",
"Cs",
"Hg"
],
"n_elements": 26,
"elements_per_component": {
"protein": {
"n": 22,
"symbols": [
"H",
"Li",
"C",
"N",
"O",
"Na",
"Mg",
"P",
"S",
"K",
"Ca",
"Mn",
"Fe",
"Co",
"Ni",
"Cu",
"Zn",
"Se",
"Sr",
"Cd",
"Cs",
"Hg"
]
},
"pocket": {
"n": 18,
"symbols": [
"H",
"C",
"N",
"O",
"Na",
"Mg",
"S",
"K",
"Ca",
"Mn",
"Fe",
"Co",
"Ni",
"Cu",
"Zn",
"Sr",
"Cd",
"Cs"
]
},
"ligand": {
"n": 9,
"symbols": [
"C",
"N",
"O",
"F",
"P",
"S",
"Cl",
"Br",
"I"
]
}
},
"vocab_sizes": {
"res_name": 50,
"chain": 33,
"insertion_code": 21,
"hetero": 30
},
"residue_identity": {
"res_idx": "derived key, contiguous 0..R-1 over protein+ligand per complex; the pocket INHERITS the protein numbering, so pocket res_idx is a subset, neither contiguous nor disjoint",
"merged_residue_complexes": {
"protein": 0,
"pocket": 0
},
"total_residues": 3311198,
"residues_per_complex": {
"min": 89,
"mean": 741.922,
"max": 9102
},
"pocket_residues_total": 122736,
"pocket_res_idx_subset_of_protein": true,
"author_key_3part_collisions": {
"protein": {
"colliding_keys": 5,
"complexes": 5,
"pdb_ids": [
"3mi2",
"3upk",
"3v4t",
"4w9f",
"4w9h"
]
},
"pocket": {
"colliding_keys": 0,
"complexes": 0,
"pdb_ids": []
}
},
"author_key_4part_collisions": {
"protein": {
"colliding_keys": 0,
"complexes": 0,
"pdb_ids": []
},
"pocket": {
"colliding_keys": 0,
"complexes": 0,
"pdb_ids": []
}
},
"res_seq": {
"protein": {
"min": -20,
"max": 5069,
"atoms_le_0": 18665,
"complexes_with_le_0": 313
},
"pocket": {
"min": -4,
"max": 4027,
"atoms_le_0": 197,
"complexes_with_le_0": 8
},
"ligand": {
"min": 0,
"max": 0,
"atoms_le_0": 113493,
"complexes_with_le_0": 4463
}
},
"insertion_code": {
"vocab_size": 21,
"vocab": [
" ",
"C",
"B",
"A",
"D",
"E",
"F",
"G",
"H",
"I",
"J",
"K",
"L",
"M",
"P",
"W",
"N",
"Q",
"R",
"S",
"T"
],
"protein_complexes_using_one": 372,
"pocket_complexes_using_one": 212,
"protein_atoms_with_one": 86164,
"pocket_atoms_with_one": 9701,
"distinct_non_blank": 20,
"water_artefact": {
"atoms": 7866,
"complexes": 5,
"note": "NOT author insertion codes: in 5 complexes (2qwb..2qwf) ATOM3D spilled the water's chain id into this column; those waters read chain == '.'. Shipped verbatim. 15 letters are involved (E F G H J K L M N P Q R S T W); the genuine author insertion codes are the non-water rows."
},
"protein_atoms_non_water_with_one": 78298
},
"invariants_over_all_complexes": {
"protein_plus_ligand_is_0..R-1_per_complex": true,
"ligand_is_one_pseudo_residue_at_R-1": true,
"pocket_res_idx_subset_of_protein": true,
"pocket_residues": 122736,
"pocket_residues_truncated": 0
}
},
"hetero_counts": {
"protein": {
" ": 26375178,
"W": 1384783,
"H_MSE": 1729,
"H_ZN": 1360,
"H_PTR": 1250,
"H_CA": 974,
"H_MG": 770,
"H_TPO": 765,
"H_SEP": 644,
"H_CSO": 513,
"H_NA": 433,
"H_CD": 202,
"H_LLP": 176,
"H_MN": 156,
"H_PCA": 156,
"H_K": 96,
"H_NI": 78,
"H_FE": 67,
"H_KCX": 65,
"H_CAS": 61,
"H_HG": 48,
"H_CO": 48,
"H_ACE": 45,
"H_FE2": 22,
"H_CSD": 8,
"H_LI": 6,
"H_CU": 6,
"H_CS": 6,
"H_SR": 2,
"H_IAS": 2
},
"pocket": {
" ": 1657566,
"H_ZN": 686,
"H_MG": 177,
"H_CA": 61,
"H_MN": 59,
"H_NA": 26,
"H_CO": 26,
"H_FE": 21,
"H_NI": 8,
"H_FE2": 6,
"H_CU": 4,
"H_K": 3,
"H_CD": 1,
"H_SR": 1,
"H_CS": 1
},
"ligand": {
" ": 113493
}
},
"water": {
"hetero_W_atoms": 1384783,
"res_name_HOH_atoms": 1384783,
"identical": true,
"note": "waters live in the protein component; the pocket has none"
},
"entity_type_counts": {
"protein": {
"protein": 26375178,
"water": 1384783,
"metal_ion": 4333,
"other_hetero": 5355
},
"pocket": {
"protein": 1657566,
"metal_ion": 1080
},
"ligand": {
"ligand": 113493
}
},
"hydrogens": {
"protein": {
"n_hydrogen": 11328810,
"fraction": 0.408,
"complexes_with_zero_H": 0
},
"pocket": {
"n_hydrogen": 682255,
"fraction": 0.4113,
"complexes_with_zero_H": 0
},
"ligand": {
"n_hydrogen": 0,
"fraction": 0.0,
"complexes_with_zero_H": 4463
}
},
"bfactor": {
"protein": {
"min": -9.37,
"mean": 16.987,
"max": 500.54,
"n_atoms": 27769649,
"n_zero": 11329695,
"n_negative": 201,
"n_hydrogen": 11328810,
"n_hydrogen_zero": 11328810,
"n_heavy_zero": 885
},
"pocket": {
"min": -0.01,
"mean": 13.3213,
"max": 500.0,
"n_atoms": 1658646,
"n_zero": 682380,
"n_negative": 27,
"n_hydrogen": 682255,
"n_hydrogen_zero": 682255,
"n_heavy_zero": 125
},
"ligand": {
"min": 0.0,
"mean": 0.0,
"max": 0.0,
"n_atoms": 113493,
"n_zero": 113493,
"n_negative": 0,
"n_hydrogen": 0,
"n_hydrogen_zero": 0,
"n_heavy_zero": 113493
}
},
"occupancy": {
"protein": {
"min": -0.49,
"mean": 0.9959,
"max": 1.99,
"n_distinct": 117,
"n_not_one": 238151,
"n_zero": 9239,
"n_below_0": 2,
"n_above_1": 22
},
"pocket": {
"min": 0.0,
"mean": 0.9954,
"max": 1.21,
"n_distinct": 98,
"n_not_one": 16451,
"n_zero": 298,
"n_below_0": 0,
"n_above_1": 1
},
"ligand": {
"min": 1.0,
"mean": 1.0,
"max": 1.0,
"n_distinct": 1,
"n_not_one": 0,
"n_zero": 0,
"n_below_0": 0,
"n_above_1": 0
}
},
"is_alpha_carbon": {
"protein": {
"n_true": 1917382,
"z_values": [
6
]
},
"pocket": {
"n_true": 121656,
"z_values": [
6
]
},
"ligand": {
"n_true": 0,
"z_values": []
}
},
"label_neglog_aff": {
"min": 2.0,
"mean": 6.454,
"max": 11.92
},
"splits": {
"seq-id-30": {
"train": 3507,
"val": 466,
"test": 490,
"total": 4463
},
"seq-id-60": {
"train": 3563,
"val": 448,
"test": 452,
"total": 4463
}
},
"ligand_bonds": {
"total": 121187,
"dtype": "torch.float32",
"order_counts": {
"1.0": 89491,
"2.0": 31543,
"3.0": 153
},
"convention": "one entry per bond (not both directions), as in HiQBind"
},
"file_sizes_gib": {
"complexes.pt": 0.079,
"protein.pt": 1.19
},
"units": {
"pos": "angstrom",
"bfactor": "angstrom^2 (B-factor)",
"occupancy": "fraction, nominally [0,1] but NOT clamped -- the source states -0.49 .. 1.99 and it is shipped verbatim",
"neglog_aff": "unitless (-log10 of the dissociation constant in M; pKd-like)"
},
"sampled_fields": "none in this block -- everything here is exact over all 4,463 complexes. The TOP-LEVEL atoms_per_structure / bounding_box_extent_angstrom / elements keys describe the pocket+ligand view only (all 4,463 complexes, so exhaustive for that view but NOT for the dataset: the protein component adds 3 elements).",
"v1_regression": {
"bitwise_identical": [
"ligand_edge_index",
"ligand_edge_ptr",
"ligand_pos",
"ligand_ptr",
"ligand_z",
"neglog_aff",
"pocket_pos",
"pocket_ptr",
"pocket_z",
"protein_pos",
"protein_ptr",
"protein_z",
"seqid30_split",
"seqid60_split"
],
"differs": [],
"ligand_edge_attr": {
"v1_dtype": "torch.int64",
"v2_dtype": "torch.float32",
"values_equal": true
},
"v1_merged_residue_complexes": {
"protein": 376,
"pocket": 208
},
"v2_merged_residue_complexes": {
"protein": 0,
"pocket": 0
},
"is_alpha_carbon": {
"protein": {
"v1_true": 1918415,
"v2_true": 1917382,
"lost": 1033,
"gained": 0,
"z_of_lost": [
20
],
"lost_residues": 1033
},
"pocket": {
"v1_true": 121717,
"v2_true": 121656,
"lost": 61,
"gained": 0,
"z_of_lost": [
20
],
"lost_residues": 61
}
},
"v1_missing_fields": [
"hetero",
"insertion_code",
"n_residues",
"occupancy",
"res_seq"
]
}
}
}