Merge pull request #33 from OpenDrugDiscovery/tmqm_add

Merging tmqm branch

README.md

@@ -2,16 +2,15 @@
 
 Open Quantum Data Commons
 
-## Setup Datasets
-
-Use the scripts in `setup/` to download the datasets. For more information, see the [README](setup/README.md) in the `setup/` directory.
-
-# Install the library in dev mode
+### Installing openQDC
+```bash
+git clone [email protected]:OpenDrugDiscovery/openQDC.git
+cd openQDC
+# use mamba/conda
+mamba env create -n openqdc -f env.yml
 pip install -e .
 ```
 
-## Development lifecycle
-
 ### Tests
 
 You can run tests locally with:

env.yml

@@ -8,39 +8,22 @@ dependencies:
   - tqdm
   - loguru
   - fsspec
-  - s3fs
   - gcsfs
-  - joblib
-  - prettytable
-  - pyrootutils
 
   # Scientific
   - pandas
   - numpy
-  - scipy
-  - sympy
 
   # Chem
-  - ipdb
   - datamol #==0.9.0
   - rdkit #-pypi #==2022.9.3
   - ase
-
-  # ML
-  #- einops =0.6.0
-  - pytorch
   - dscribe
 
   # other stuffs
   - h5py >=3.8.0
   - gdown #==4.6.4
 
-  # Viz
-  - matplotlib
-  - seaborn
-  - ipywidgets
-  - nglview
-
   # Dev
   - pytest >=6.0
   - pytest-cov
@@ -49,8 +32,6 @@ dependencies:
   - jupyterlab
   - pre-commit
   - ruff
-  - ipykernel
-  - isort
 
   # Doc
   - mkdocs

src/openqdc/datasets/base.py

@@ -6,7 +6,6 @@
 
 import numpy as np
 import pandas as pd
-import torch
 from ase.io.extxyz import write_extxyz
 from loguru import logger
 from sklearn.utils import Bunch
@@ -82,7 +81,7 @@ def read_qc_archive_h5(
     return samples
 
 
-class BaseDataset(torch.utils.data.Dataset):
+class BaseDataset:
     __energy_methods__ = []
     __force_methods__ = []
     energy_target_names = []
@@ -143,10 +142,18 @@ def _precompute_statistics(self, overwrite_local_cache: bool = False):
             logger.info("Loaded precomputed statistics")
         else:
             logger.info("Precomputing relevant statistics")
-            (formation_E_mean, formation_E_std, total_E_mean, total_E_std) = self._precompute_E()
+            (
+                inter_E_mean,
+                inter_E_std,
+                formation_E_mean,
+                formation_E_std,
+                total_E_mean,
+                total_E_std,
+            ) = self._precompute_E()
             forces_dict = self._precompute_F()
             stats = {
                 "formation": {"energy": {"mean": formation_E_mean, "std": formation_E_std}, "forces": forces_dict},
+                "inter": {"energy": {"mean": inter_E_mean, "std": inter_E_std}, "forces": forces_dict},
                 "total": {"energy": {"mean": total_E_mean, "std": total_E_std}, "forces": forces_dict},
             }
             with open(local_path, "wb") as f:
@@ -162,20 +169,24 @@ def _precompute_E(self):
         s = np.array(self.data["atomic_inputs"][:, :2], dtype=int)
         s[:, 1] += IsolatedAtomEnergyFactory.max_charge
         matrixs = [matrix[s[:, 0], s[:, 1]] for matrix in self.__isolated_atom_energies__]
-        converted_energy_data = self.convert_energy(self.data["energies"])
+        converted_energy_data = self.data["energies"]
         # calculation per molecule formation energy statistics
         E = []
         for i, matrix in enumerate(matrixs):
             c = np.cumsum(np.append([0], matrix))[splits_idx]
             c[1:] = c[1:] - c[:-1]
             E.append(converted_energy_data[:, i] - c)
         E = np.array(E).T
+        inter_E_mean = np.nanmean(E / self.data["n_atoms"][:, None], axis=0)
+        inter_E_std = np.nanstd(E / self.data["n_atoms"][:, None], axis=0)
         formation_E_mean = np.nanmean(E, axis=0)
         formation_E_std = np.nanstd(E, axis=0)
         total_E_mean = np.nanmean(converted_energy_data, axis=0)
         total_E_std = np.nanstd(converted_energy_data, axis=0)
 
         return (
+            np.atleast_2d(inter_E_mean),
+            np.atleast_2d(inter_E_std),
             np.atleast_2d(formation_E_mean),
             np.atleast_2d(formation_E_std),
             np.atleast_2d(total_E_mean),
@@ -253,12 +264,25 @@ def data_shapes(self):
             "forces": (-1, 3, len(self.force_target_names)),
         }
 
+    @property
+    def atoms_per_molecules(self):
+        try:
+            if hasattr(self, "_n_atoms"):
+                return self._n_atoms
+            self._n_atoms = self.data["n_atoms"]
+            return self._n_atoms
+        except:  # noqa
+            return None
+
     def _set_units(self, en, ds):
         old_en, old_ds = self.energy_unit, self.distance_unit
-        if en is not None:
-            self.set_energy_unit(en)
-        if ds is not None:
-            self.set_distance_unit(ds)
+        en = en if en is not None else old_en
+        ds = ds if ds is not None else old_ds
+
+        # if en is None:
+        self.set_energy_unit(en)
+        # if ds is not None:
+        self.set_distance_unit(ds)
         if self.__force_methods__:
             self.__forces_unit__ = self.energy_unit + "/" + self.distance_unit
             self.__class__.__fn_forces__ = get_conversion(old_en + "/" + old_ds, self.__forces_unit__)
@@ -556,7 +580,7 @@ def get_statistics(self, normalization: str = "formation", return_none: bool = T
         """
         Get the statistics of the dataset.
         normalization : str, optional
-            Type of energy, by default "formation", must be one of ["formation", "total"]
+            Type of energy, by default "formation", must be one of ["formation", "total", "inter"]
         return_none : bool, optional
             Whether to return None if the statistics for the forces are not available, by default True
             Otherwise, the statistics for the forces are set to 0.0

src/openqdc/datasets/dummy.py

@@ -23,22 +23,22 @@ class Dummy(BaseDataset):
 
     force_target_names = [f"forces{i}" for i in range(len(__force_methods__))]
     __isolated_atom_energies__ = []
-    __average_n_atoms__ = 20
+    __average_n_atoms__ = None
 
     @property
     def _stats(self):
         return {
             "formation": {
                 "energy": {
-                    "mean": array([-12.94348027, -9.83037297]),
-                    "std": array([4.39971409, 3.3574188]),
+                    "mean": array([[-12.94348027, -9.83037297]]),
+                    "std": array([[4.39971409, 3.3574188]]),
                 },
                 "forces": NOT_DEFINED,
             },
             "total": {
                 "energy": {
-                    "mean": array([-89.44242, -1740.5336]),
-                    "std": array([29.599571, 791.48663]),
+                    "mean": array([[-89.44242, -1740.5336]]),
+                    "std": array([[29.599571, 791.48663]]),
                 },
                 "forces": NOT_DEFINED,
             },
@@ -51,6 +51,11 @@ def __init__(self, energy_unit=None, distance_unit=None, cache_dir=None) -> None
         except:  # noqa
             pass
         self._set_isolated_atom_energies()
+        self.setup_dummy()
+
+    def setup_dummy(self):
+        self._n_atoms = np.array([np.random.randint(1, 100) for _ in range(self.__len__())])
+        self.__average_nb_atoms__ = self._n_atoms.mean()
 
     def is_preprocessed(self):
         return True
@@ -63,7 +68,7 @@ def __len__(self):
 
     def __getitem__(self, idx: int):
         shift = IsolatedAtomEnergyFactory.max_charge
-        size = np.random.randint(1, 100)
+        size = self._n_atoms[idx]
         z = np.random.randint(1, 100, size)
         c = np.random.randint(-1, 2, size)
         return Bunch(

src/openqdc/datasets/pcqm.py

@@ -73,9 +73,6 @@ class PCQM_PM6(BaseDataset):
     __force_methods__ = []
     force_target_names = []
 
-    def __init__(self, energy_unit=None, distance_unit=None) -> None:
-        super().__init__(energy_unit=energy_unit, distance_unit=distance_unit)
-
     @property
     def root(self):
         return p_join(get_local_cache(), "pubchemqc")
@@ -161,8 +158,4 @@ def collate_and_save_list(self, list_entries):
 class PCQM_B3LYP(PCQM_PM6):
     __name__ = "pubchemqc_b3lyp"
     __energy_methods__ = ["b3lyp"]
-
     energy_target_names = ["b3lyp"]
-
-    def __init__(self, energy_unit=None, distance_unit=None) -> None:
-        super().__init__(energy_unit=energy_unit, distance_unit=distance_unit)

src/openqdc/datasets/spice.py

@@ -13,7 +13,7 @@ def read_record(r):
     smiles = r["smiles"].asstr()[0]
     subset = r["subset"][0].decode("utf-8")
     n_confs = r["conformations"].shape[0]
-    x = get_atomic_number_and_charge(dm.to_mol(smiles, add_hs=True))
+    x = get_atomic_number_and_charge(dm.to_mol(smiles, remove_hs=False, ordered=True))
     positions = r["conformations"][:]
 
     res = dict(

src/openqdc/datasets/transition1x.py

@@ -55,6 +55,10 @@ class Transition1X(BaseDataset):
         "wB97x_6-31G(d).forces",
     ]
 
+    __energy_unit__ = "hartree"
+    __distance_unit__ = "ang"
+    __forces_unit__ = "hartree/ang"
+
     def read_raw_entries(self):
         raw_path = p_join(self.root, "Transition1x.h5")
         f = load_hdf5_file(raw_path)["data"]

src/openqdc/utils/__init__.py

@@ -6,7 +6,6 @@
     load_hdf5_file,
     load_json,
     load_pkl,
-    load_torch,
     makedirs,
     save_pkl,
     set_cache_dir,
@@ -19,7 +18,6 @@
     "makedirs",
     "load_hdf5_file",
     "load_json",
-    "load_torch",
     "create_hdf5_file",
     "check_file",
     "set_cache_dir",