notebooks/tutorial_custom_moleculenet.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "# Moleculenet"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "The `MoleculeNetDataset` class is intended for datasets that consist of a table of smiles and corresponding targets and convert them into a tensors representation for graph networks. The class provides properties and methods for making graph features from smiles.\n",
    "The typical input is a `csv` or `excel` file with smiles and corresponding graph labels.\n",
    "\n",
    "The graph structure matches the molecular graph, i.e. the chemical structure. And features for atoms and bonds are generated with `RDkit` chemical informatics software.\n",
    "\n",
    "The atomic coordinates\n",
    "are generated by a conformer guess. Since this require some computation time, it is only done once and the\n",
    "molecular coordinate or mol-blocks stored in a single SDF file with the base-name of the csv file."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "For demonstration, we make an artifical table of smiles and some values and store them to file."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "import os\n",
    "os.makedirs(\"ExampleMol\", exist_ok=True)\n",
    "csv_data = \"\".join([\n",
    "    \"smiles,Values1,Values2\\n\",  # Need header!\n",
    "    \"CCC, 1, 0.1\\n\",\n",
    "    \"CCCO, 2, 0.3\\nCCCN, 3, 0.2\\n\",\n",
    "    \"CCCC=O, 4, 0.4\\n\"\n",
    "    \"NOCF, 4, 1.4\\n\"\n",
    "])\n",
    "with open(\"ExampleMol/data.csv\", \"w\") as f:\n",
    "    f.write(csv_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "The file structure is:\n",
    "\n",
    "\n",
    "```bash\n",
    "├── ExampleMol\n",
    "    ├── data.csv\n",
    "    └── data.sdf  # After prepare_data\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "from kgcnn.data.moleculenet import MoleculeNetDataset, OneHotEncoder"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "## 1. Initialization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "To load the dataset from memeory the ``MoleculeNetDataset`` class requires the information of the directory the data is in and the name of the csv-file. Also recommended is to provide a name of the dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "dts = MoleculeNetDataset(file_name=\"data.csv\", \n",
    "                         data_directory=\"ExampleMol/\", \n",
    "                         dataset_name=\"ExampleMol\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "## 2. Data Preparation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "Precompute the molecular structure and possibly also coordinates and cache the information to file as SDF mol table format in the same folder as provided in the class initialization. The structure generation can be run in parallel but the SDF file generated may be large and must still fit in memory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:kgcnn.data.ExampleMol:Generating molecules and store ExampleMol/data.sdf to disk...\n",
      "INFO:kgcnn.data.ExampleMol: ... converted molecules 5 from 5\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<MoleculeNetDataset []>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dts.prepare_data(\n",
    "    overwrite=True, \n",
    "    smiles_column_name=\"smiles\", \n",
    "    add_hydrogen=True,\n",
    "    make_conformers=True,\n",
    "    optimize_conformer=True,\n",
    "    num_workers=None  # Default is #cpus\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "## 3. Read Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "After ``prepare_data()`` is called, the cached mol-file can be read directly from the data-directory.\n",
    "The reading step can also define the labels or targets to assigning property `graph_labels` from the column of the csv table. By default a simple graph is generated without attributes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:kgcnn.data.ExampleMol:Read molecules from mol-file.\n",
      "INFO:kgcnn.data.ExampleMol: ... process molecules 0 from 5\n",
      "INFO:kgcnn.molecule.encoder:OneHotEncoder Symbol found ['C', 'O', 'N', 'F']\n",
      "INFO:kgcnn.molecule.encoder:OneHotEncoder Hybridization found [rdkit.Chem.rdchem.HybridizationType.SP3, rdkit.Chem.rdchem.HybridizationType.SP2]\n",
      "INFO:kgcnn.molecule.encoder:OneHotEncoder TotalDegree found [4, 2, 3, 1]\n",
      "INFO:kgcnn.molecule.encoder:OneHotEncoder TotalNumHs found [3, 2, 1, 0]\n",
      "INFO:kgcnn.molecule.encoder:OneHotEncoder CIPCode found [None]\n",
      "INFO:kgcnn.molecule.encoder:OneHotEncoder ChiralityPossible found [None]\n",
      "INFO:kgcnn.molecule.encoder:OneHotEncoder BondType found [rdkit.Chem.rdchem.BondType.SINGLE, rdkit.Chem.rdchem.BondType.DOUBLE]\n",
      "INFO:kgcnn.molecule.encoder:OneHotEncoder Stereo found [rdkit.Chem.rdchem.BondStereo.STEREONONE]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of graphs: 5\n"
     ]
    }
   ],
   "source": [
    "dts.read_in_memory(\n",
    "    label_column_name=[\"Values1\", \"Values2\"], \n",
    "    add_hydrogen=False,  # We remove H's \n",
    "    has_conformers=True  # We keep strucutre\n",
    ")\n",
    "print(\"Number of graphs:\", len(dts))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "## 4. Setting Attributes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "For molecular graphs nodes, edges or atoms and bonds should have attributes that are for `MoleculeNetDataset` generated from `RDkit` that includes chirality, stereo information (etc). Note that if a smile can not be processed by `RDkit` the graph will not have attributes.\n",
    "\n",
    "This can be achieved by setting a list of identifiers of predefined attributes or supply custom functions.\n",
    "Additionally an encoder can be provided to cast or transform the `RDkit` data formats into a list or value that eventually be cast into numpy `dtype=\"float\"` array. Also one-hot-encoding or mapping on to distributions can be handled by encoders. Attributes "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Atoms: ['NumBonds', 'AtomicNum', 'AtomMapNum', 'Idx', 'Degree', 'TotalDegree', 'Symbol', 'NumExplicitHs', 'NumImplicitHs', 'TotalNumHs', 'IsAromatic', 'Isotope', 'TotalValence', 'Mass', 'MassScaled', 'IsInRing', 'Hybridization', 'NoImplicit', 'ChiralTag', 'FormalCharge', 'ExplicitValence', 'ImplicitValence', 'NumRadicalElectrons', 'HasOwningMol', 'PDBResidueInfo', 'MonomerInfo', 'Smarts', 'CIPCode', 'CIPRank', 'ChiralityPossible', 'MolFileRLabel', 'GasteigerCharge', 'GasteigerHCharge', 'AtomFeatures', 'DescribeQuery', 'Rvdw', 'RvdwScaled', 'Rcovalent', 'RcovalentScaled']\n",
      "Bonds: ['BondType', 'IsAromatic', 'IsConjugated', 'IsInRing', 'Stereo', 'Idx', 'BeginAtom', 'BeginAtomIdx', 'BondDir', 'BondTypeAsDouble', 'EndAtom', 'EndAtomIdx', 'Smarts', 'DescribeQuery']\n",
      "Molecule: ['C', 'N', 'O', 'H', 'S', 'F', 'Cl', 'NumAtoms', 'AtomsIsInRing', 'AtomsIsAromatic', 'NumBonds', 'BondsIsConjugated', 'BondsIsAromatic', 'NumRotatableBonds', 'ExactMolWt', 'FpDensityMorgan3', 'FractionCSP3', 'MolLogP', 'MolMR', 'fr_Al_COO', 'fr_Ar_COO', 'fr_Al_OH', 'fr_Ar_OH', 'fr_C_O_noCOO', 'fr_NH2', 'fr_SH', 'fr_sulfide', 'fr_alkyl_halide']\n"
     ]
    }
   ],
   "source": [
    "# Class to make attributes used by MoleculeNetDataset\n",
    "import rdkit.Chem as Chem\n",
    "from kgcnn.molecule.graph_rdkit import MolecularGraphRDKit\n",
    "mol = MolecularGraphRDKit()\n",
    "# Identifiers:\n",
    "print(\"Atoms:\", list(mol.atom_fun_dict.keys()))\n",
    "print(\"Bonds:\", list(mol.bond_fun_dict.keys()))\n",
    "print(\"Molecule:\", list(mol.mol_fun_dict.keys()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "Custom functions must take a `RDkit` Mol, Atom or Bond instance as input for node edge or graph respectively."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# Or make custom function\n",
    "def mol_feature(m):\n",
    "    return m.GetNumAtoms()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "Or using a callback directly for a new attribute, which takes the csv table and molecule list as argument."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "def graph_size_callback(mg, ds):\n",
    "    return mg.mol.GetNumAtoms()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "Or using custom transform"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "def custum_trafo(mg):\n",
    "    return mg.compute_partial_charges()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:kgcnn.data.ExampleMol:Read molecules from mol-file.\n",
      "INFO:kgcnn.data.ExampleMol: ... process molecules 0 from 5\n",
      "INFO:kgcnn.molecule.encoder:OneHotEncoder Symbol found ['C', 'O', 'N', 'F']\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<MoleculeNetDataset [{'node_symbol': array(['C', 'C', 'C'], dtype='<U1'), 'node_number': array([6, 6, 6]), 'edge_indices': array([[0, 1],\n",
       "       [1, 0],\n",
       "       [1, 2],\n",
       "       [2, 1]], dtype=int64), 'edge_number': array([1, 1, 1, 1]), 'graph_size': array(3), 'node_coordinates': array([[ 1.2106, -0.1633, -0.3777],\n",
       "       [ 0.0398, -0.1833,  0.5905],\n",
       "       [-1.2407,  0.3016, -0.0679]]), 'graph_labels': array([1, 0.1], dtype=object), 'node_attributes': array([[ 1.        ,  0.        ,  0.        ,  3.        , -0.06564544],\n",
       "       [ 1.        ,  0.        ,  0.        ,  2.        , -0.05903836],\n",
       "       [ 1.        ,  0.        ,  0.        ,  3.        , -0.06564544]],\n",
       "      dtype=float32), 'edge_attributes': array([[1., 0.],\n",
       "       [1., 0.],\n",
       "       [1., 0.],\n",
       "       [1., 0.]], dtype=float32), 'graph_attributes': array([44.0626,  3.    ], dtype=float32), 'size': array(3)} ...]>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dts.set_attributes(\n",
    "    # Nodes\n",
    "    nodes=[\"Symbol\", \"TotalNumHs\", \"GasteigerCharge\"], \n",
    "    encoder_nodes={\n",
    "        \"Symbol\": OneHotEncoder([\"C\", \"N\", \"O\"], dtype=\"str\", add_unknown=False)\n",
    "    },\n",
    "    # Edges\n",
    "    edges=[\"BondType\", \"Stereo\"], \n",
    "    encoder_edges = {\n",
    "        \"BondType\": int\n",
    "    },\n",
    "    # Graph-level\n",
    "    graph=[\"ExactMolWt\", mol_feature],\n",
    "    additional_callbacks= {\"size\": graph_size_callback},\n",
    "    custom_transform=custum_trafo\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "## 4. Checking graphs in dataset "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "import networkx as nx"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([array([6, 6, 6]),\n",
       "  array([6, 6, 6, 8]),\n",
       "  array([6, 6, 6, 7]),\n",
       "  array([6, 6, 6, 6, 8]),\n",
       "  array([7, 8, 6, 9])],\n",
       " [array(['C', 'C', 'C'], dtype='<U1'),\n",
       "  array(['C', 'C', 'C', 'O'], dtype='<U1'),\n",
       "  array(['C', 'C', 'C', 'N'], dtype='<U1'),\n",
       "  array(['C', 'C', 'C', 'C', 'O'], dtype='<U1'),\n",
       "  array(['N', 'O', 'C', 'F'], dtype='<U1')])"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dts.get(\"node_number\"), dts.get(\"node_symbol\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'node_symbol': array(['C', 'C', 'C', 'C', 'O'], dtype='<U1'), 'node_number': array([6, 6, 6, 6, 8]), 'edge_indices': array([[0, 1],\n",
      "       [1, 0],\n",
      "       [1, 2],\n",
      "       [2, 1],\n",
      "       [2, 3],\n",
      "       [3, 2],\n",
      "       [3, 4],\n",
      "       [4, 3]], dtype=int64), 'edge_number': array([1, 1, 1, 1, 1, 1, 2, 2]), 'graph_size': array(5), 'node_coordinates': array([[-1.4031,  0.6186,  0.2944],\n",
      "       [-0.5738, -0.2774, -0.6146],\n",
      "       [ 0.6892, -0.8024,  0.059 ],\n",
      "       [ 1.7448,  0.2668,  0.1729],\n",
      "       [ 2.8552,  0.1743, -0.3415]]), 'graph_labels': array([4, 0.4], dtype=object), 'node_attributes': array([[ 1.0000000e+00,  0.0000000e+00,  0.0000000e+00,  3.0000000e+00,\n",
      "        -6.4970836e-02],\n",
      "       [ 1.0000000e+00,  0.0000000e+00,  0.0000000e+00,  2.0000000e+00,\n",
      "        -4.9520925e-02],\n",
      "       [ 1.0000000e+00,  0.0000000e+00,  0.0000000e+00,  2.0000000e+00,\n",
      "         2.7463208e-03],\n",
      "       [ 1.0000000e+00,  0.0000000e+00,  0.0000000e+00,  1.0000000e+00,\n",
      "         1.1947978e-01],\n",
      "       [ 0.0000000e+00,  0.0000000e+00,  1.0000000e+00,  0.0000000e+00,\n",
      "        -3.0340311e-01]], dtype=float32), 'edge_attributes': array([[1., 0.],\n",
      "       [1., 0.],\n",
      "       [1., 0.],\n",
      "       [1., 0.],\n",
      "       [1., 0.],\n",
      "       [1., 0.],\n",
      "       [2., 0.],\n",
      "       [2., 0.]], dtype=float32), 'graph_attributes': array([72.05752,  5.     ], dtype=float32), 'size': array(5)}\n"
     ]
    }
   ],
   "source": [
    "print(dts[3])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "G = nx.Graph()\n",
    "G.add_nodes_from([(i, {\"atom\": x}) for i, x in enumerate(dts.get(\"node_symbol\")[3])])\n",
    "G.add_edges_from(dts.get(\"edge_indices\")[3])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "labels = nx.get_node_attributes(G, 'atom') \n",
    "nx.draw(G,labels=labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAcIAAACWCAIAAADCEh9HAAAABmJLR0QA/wD/AP+gvaeTAAASY0lEQVR4nO3da1BU5xkH8Ie74A0XFFDwgmAgKJdoMIkaL5CASj9kLJP6YTP1Q+lMx2Fsmglpp8lqZzJDTW23maYt5kNnU5uqM2EyGzVWUIgagyKQaAQFNALiBRQwIArIvv3wrrvLgsDue86ec3b/v/HDkeye80bO/vdc3uc5fowxAgAAd/krPQAAAG1DjAIACEGMAgAIQYwCAAhBjAIACEGMAgAIQYwCAAhBjAIACEGMAgAIQYwCAAhBjAIACEGMAgAIQYwCAAhBjAIACEGMAgAIQYwCAAhBjAIACEGMAgAIQYwCAAhBjAIACEGMAgAIQYwCAAhBjAIACEGMAgAIQYwCAAhBjAIACEGMAgAIQYwCAAhBjAIACEGMAgAIQYwCAAhBjAIACEGMAgAIQYx6wpEjR9LT09PT0z/99FOlxwJeoqmpKTMzMyMj45133lF6LL7OjzGm9Bi8WUtLS35+fnV1te0nCQkJpaWly5YtU3BUoGmPHj3atm3bgQMHbB/e2NjYw4cPp6amKjswn4WjUbkMDQ1t27YtPj6eZ2hCQsKKFSuIqLm5OS0t7dVXX+3t7VV6jKA9u3fvDg8P379/P2NMp9Pl5ub6+/vfuHEjIyPjjTfe6OjoUHqAPomBDCoqKhITE/m/sG2nZ4wdO3YsJiaG/zwmJubgwYPKjhM0pKmp6bXXXuM7T1BQ0Ntvv81/fuPGjaKiouDgYCKaNWtWcXHxwMCAskP1NYhRibW1ten1ej8/PyKaPn369u3bLRaL02uKi4t1Oh3/PGRlZV24cEGRoYJW9PX1GQyGkJAQIgoJCcnOzv7xxx+dXnP58uVNmzbxnWrJkiWHDh1SZKi+CTEqmQcPHhgMhilTphBRWFiYwWDo7+9/2ouHhoZKSkoiIyOJKDAwsKCgoLOz05OjBU2wWCwmkyk6OpqI/Pz89Hr9rVu3xnl9WVnZs88+y8M0Ozv70qVLHhuqL0OMSsBisRw8eHDBggV8X8/Pz79+/fpk3njv3r3CwsLAwEAi0ul0RqNxaGhI7tGCVpw9e/aFF17gmZiZmXnmzJnJvGtwcNBoNM6cOZOf+xcWFvb09Mg9VB+HGBV1/vz5VatW8X19+fLlp0+fdnUNDQ0Nubm5fA1JSUlffvmlHOMEDblx44bt0tC8efNMJtPoS0Pju3v3bmFhYUBAABFFREQYjcbHjx/LNFpAjLrv5s2bBQUF/v7+/H5RSUnJ8PCw22szm83x8fE8TPPy8pqbmyUcKmhFf39/cXHxtGnTiCg0NLSoqKi3t9fttdXU1KxZs4bvVBkZGV999ZWEQwUbxKg7BgYGjEbjjBkziCg4OLiwsPD+/fvSrpafjkmyWtAKs9m8cOFC21fpDz/8oObVgg1i1GVOh41Xr16Vdv3SHuSCJjgdNp48eVLa9Y8+yB19rx/chhh1gScvYjpdcj116pR82wIFefIipvglVxgTYnRSFLml7vYEANAEpW6pnz179sUXX+Tf0M8///wkJwDAOBCjE1B8giefjhoaGjqZ6aigFcpO8HR1OiqMDzE6nvLy8qVLl9rKjS5evKjUSByLo2JjY3E6pl2O5UbPPPPM4cOHlRqJY3HU1KlTDQbDw4cPlRqMpiFGx9bY2Jifn8/39YSEBJUUv1dWVqalpfFRrV27tq6uTukRgQu6urpUWPze1NSkwl1dWxCjznp7e52+oh89eqT0oOyGh4dNJlNUVBQR+fv76/X627dvKz0omAD/rc2ZM8f2W7tz547Sgxrh+PHjtuaNGzZsQJ8HlyBG7TSUUN3d3UVFRTzrw8PDi4uLVZX14OjEiRO2TqDr16//7rvvlB7R2PhtgNmzZ9tuA3R0dCg9KG1AjFpVVVU51i9/8803So9oYleuXMnLy+NjTkxMxOmY2rS2tur1ev4LiouLM5lMSo9oYl1dXbZJKbNmzUKfh8lAjI64e6PFyXRlZWUpKSm2e77ff/+90iMC690b3u5Li3dvGhoaNm7cqIb7YJrg0zH64MEDW2lHWFiYYP2ygvgMxPDwcD4DEW33FMRn+8bFxdlm+7a0tMi3uU8++WTv3r0y1bmZzebFixfbvqHr6+vl2IoX8N0Y9b5CY14jwOtheI0Amvp42Llz51566SXbzPavv/5a1s319PTwGc3ytR3h39Do8zA+X4zRmpqa1atX8339ueeek7x+WVn19fU5OTn8/y45Ofno0aNKj8gntLe32zohzJ0712OdEMxm86JFi2xHA9euXZNjK7zPA/+GjoyMxDe0E9+K0c7OTtvxmnfvDU6fLsn7p4AN7/oxffp0etLuy8NdPxwHIGvbEe8+/hDhKzHqg+cmvO2egh9vX+CZg8HJcGw7MnfuXJnulFosln379sXGxhLRMxERQz//OWtrk3wrmuMTMVpWVpacnOybV8qVOtn0enV1dS+//DLfqdLT0ysrK5UeEWOMnTt3zgNtR/r6+t59991LGzYwIhYWxnbtYr7d58HLY1Q99cvKcrz1sWLFCrlvfXg3x9Z2KryV59R2JD8/v7W1VZYttbUxvZ75+TEiFhvLTCamqZmCEvLaGHWaRayS+mUF8Yk48+fP98xEHK80urVdd3e30oMam+cmrlZVsZUrGREjYpmZTAt1K5LzwhhFTds4tD4tXEFaLHNobm62tR2RsYxqeJiZTCwqihExf3+m1zO1VlHLxNti1KnDgmrrl5WlxSJFBTkW3S5ZsuSLL75QekSuOX78uCeK+nt7mcHAQkIYEZs6lRkMzGf6PHhPjDr2+1q8eDEKzCfk1DLj22+/VXpEquM1LWB42x1+isbb7sh1itbYyPLzref4iYnMNz6G3hCjOFF1m/obuCll9L+Matt9TZ7nGp6Wl7OlS61hmpXFlOt37hnajlE8C0ESjp8ufszl47fjKioqHNtje9lxuoemrwwNsZISFhnJiFhgICsoYN7b50HDMeqZKXK+4/Lly5s3b7ZdATx06JDSI1IAv2rsCw9rcZpMLdfDoO7dY4WFLCCAETGdjhmNzBvb7mkyRvGcWPko+6g1BfFHB/JLQ/zRgV5/achzjyatr2e5udZz/KQkJueTyRWhsRj1WPmwL1Pqwb9KGT2j1qceZO1YTRARESFjNYHZzOLjrWGal8e8qM+DlmLUqbWdgvXLvsBzny5FVVdXr1q1iu9Uy5cvP336tNIjUkZNTc2aNWtkbzsyMMCMRjZjBiNiwcGssJB5RWsLbcRobW2trX5ZvtaKMFptba3t0+Vl//K8+RvvNhATE4NuA8xjTXhv3mQFBczfnxGxmBhWUsI0/i+v9hj1kWMilfOy8wDe+4q3++K9r7y+3dfkOT4Sgl83k+uRENXVbNUq6zn+8uVMy+cB6o1RX7tCp3L8qrTjp0ujV6XNZnN8fLztKwGdWMfU0tLy+uuv87u4zZs3s/37ZWk7YrGwgwfZ/PmMiPn5sfx8ps2r0iqNUaf7xZqoX/YFmp4jUV9fn5uby3eqpKSkL73ufrHkTp48+actW6wHjKtXs5oaWTbz4AEzGNiUKda2ewYD09ocCdXF6JUrVzB7UeU0N2N39FOq8NDgyeJtR6KjrQeMej2TqcKltVW7bfdUFKO8fhm1NJqglfox3u6LP/eNt/vCM1Pd0dfnobYjFRUsLc16/Lt2LdNI/ZgqYhSV3Rql8m4G5eXlS5cu5UfNWVlZF729slt2TU32tiMJCXK1HeHHv3Pm2NvuqT4NlI9Rxz5D69at87L6ZV/g2FsrISFBDb21GhsbbUNKTExUw5C8R3k5W7bM3nbkwgVZttLdzYqKrMe/4eGsuFjNbfeUjFF0vfQmKun02tvbazAYeGu7adOmGQwGjba2U7XRbUdkart35QrLy7NG9pIlTK2dXpWJUceTQR+pX/YFyj53gF8aioqKIi9qbadqvO1IYKDsbUfKylhKijVMs7OZ+ubteDpG8UQgr+f0FCzP3BavqqpauXIlPxbOzMz8xiefCKSMhoYRbUeOHJFlK4ODzGhkM2cyIhYUxAoLmZqeguXRGK2ursbzKX1EQ0PDxo0b+e9a1meytrW1+UhrO1VzajvS3CzLVu7cYb/4hbWKVE3Xuz0Uo3haum8ym82LFy+2lVHU19dLuHLHssWwsDAZyxZhMhzbjvADRplKbGtr2Y4dqppVKnuM8vpl3tqO1y9rtIgQ3MOLenkBOy/qlaSA3Ww2L1iwQPYmGuAqr2s7MhnyxqjZbF60aBHql4G3U+J1RJGRkSItZmpqalavXm1r6Xbq1ClphwoSOH9+RNsRz/+O+vpYSwurqWFXrzL5j9vkitH6+vqcnBy+rycnJx89elSmDYGGCCaghFkMsuNtRxYs8FzbkZ4e9vHHLD+f6XTWBLf9mTqV5eWxDz+UqZJV+hgdXb+MfR0cuXE+LtOVAZAdbzsSGmpvO9LfL/1WHj5kH3zAIiKc03P0n6lT2e9+x6TuFSdljA4ODtrql4OCglC/DE8z+u7QOF0QP//8c/nuU4EntLXJ2Hbk3j37BYRJ/omPZ42Nkg1AwhgtKytLSUmx7euoX4YJOc5VCggIKCgocJq/cfjwYT6ZX+5ZU+AJlZUsPd3edqSuToJ13r7NkpJGRGRqKvvjH1ldHevoYBYLu3uXXbzIjEa2cuWIl0VHM+ke1yhBjDY2Nubl5aF+GdxTVVXFu9IQ0fTp03lNcHNzs63Tgr+//549e9DazhvwtiNRUfa2IyKVZhYL27TJnoyhoezjj586McBiYf/9r3UCvy1wJSoUForR4eHhN998MygoiLe227Nnz+DgoCTDAp8yNDSk1+v5tGIimj17Nj9EJaLU1NRmmeZyg1Kkajvyj3+MuOhZUTHxW86dY+Hh9ncVFbmz3VFEj0a3bt2K+mWQRHt7u60bNL8/WVpaqvSgQDaNjfa2I4mJLlclPX7MFi60B+KHH072jfv22d8VFsbu3XN14KP5McZIQHt7e2dnZ3p6ushKAGyOHTtWWloaHR29c+dOpccC8isvpx076NIlIqLsbDIa6cktlgmUltKWLdbllBS6cIGenM1MbP16qqy0Ln/wAb31lisjHoNojAIACBkaor//nXbupJ4eCgqibdvo/fcpMnKCd/30p/TZZ9blP/+Zfv1rF7Z44AD97GfW5YwMqq11fdAjIEYBQAW6umjXLvroIxoeJp2O3nuPtm+ngICnvn7ePLp507rc2kpxcS5sa2CAwsPp0SMiosBA6u6madMEhk6TPgwGAJCPTkd//StdvEg5OdTVRTt20LJl9L//jf3i69ftGTpnjmsZSkQhIfSkxTg9fkzV1e4O2goxCgCqkZxMR4+S2UyLFlFDA+3dO/bLrl2zL9sC0SVpafblq1fdWYODQMH3AwBI7Cc/oexs2rOHtm4d+wVdXfblCa+ijsnxXd3d7qzBAWIUANQnNJR+//un/teeHvvyjBnurH/mTPuycIzipB4AtObxY/tyoFvHgsHB9uXBQcHhIEYBQGscjyX7+txZw/379mWdTnA4iFEA0JrwcPuyYyBOnuO7HNfmFsQoAGjNk142RO7eZ29uti9HRQkOB9PvAUBrBgZo5kwaGCAiCgig3l4KDXVtDY6z91taaP58keHgaBQAtCYkhDIyrMvDw3TmjGtvb2qyZ2hMjGCGEmIUADQpK8u+/O9/u/Zek8m+nJ0tPhac1AOABl2/TgkJNDxMRBQWRvX19OQBXxPo6qLkZOrosP71zBlyaM/oHhyNAoAGLVxImzdbl/v76Ve/okkeEb71lj1DV6wQz1BCjAKAVv3lL/bOTEeOkF4/Ylr+mHbupH/9y7ocGEh/+5skA0GMAoA2xcfT7t32v/7nP7RuHdXUjP3ipibKy6Ndu+w/+e1vaeVKSQaCa6MAoGUGA/3hDyN+kppKOTkUF0dz5tDdu3TrFpWXU3U1WSz21/zyl/TRR+P1M3UFYhQANO6f/6Tf/Ib6+yf14qAgeu+98fqeuA4xCgDa195OO3eSyURDQ099TUAAbdlC779PCQnSbhwxCgDeoreXTpyg8nJqaaHOTrp7l3Q6mj2b5s2jDRvolVfEu5CMCTEKACAEd+oBAIQgRgEAhCBGAQCEIEYBAIQgRgEAhCBGAQCEIEYBAIQgRgEAhCBGAQCEIEYBAIQgRgEAhCBGAQCEIEYBAIQgRgEAhCBGAQCEIEYBAIQgRgEAhCBGAQCEIEYBAIQgRgEAhCBGAQCEIEYBAIQgRgEAhCBGAQCEIEYBAIQgRgEAhCBGAQCEIEYBAIQgRgEAhCBGAQCEIEYBAIQgRgEAhCBGAQCEIEYBAIT8HxMWtkBj/gsrAAAAhnpUWHRyZGtpdFBLTCByZGtpdCAyMDIzLjAzLjMAAHice79v7T0GIOBlgAAmIGYFYhYgbmBkY0gA0ozMEJqJCUFnAGlmZkYOBg0Qg4mbgZGBkYmBiZmBmYWDSQRkkrgbSCvUWAbWh25qB2bNnLkPxHnotmx/WtozO5gkkrg9TByo3gEmLgYAGj0aHMRAqn0AAADCelRYdE1PTCByZGtpdCAyMDIzLjAzLjMAAHicjVBJDoMwDLznFfMBIpOlJUcgqKoqEqml/UPv/F911IbAoQg7lryMR54IJLv723vGYsoLAdDOc87hpYlIjEgJuuFyDeintsudPj7D9ICF4Q32LbKd4pg7NXpUSlrX0PmEimQiJl6R9EsyUiVkLZVzpBueG/sHqBm4dPcYDQOPEFpEHDpxCH4j7iu3i8EXuclV0cQFdLm85jDlPsVh1+xrrlTnT+dcfABG+FetPEWPkgAAAFB6VFh0U01JTEVTIHJka2l0IDIwMjMuMDMuMwAAeJxzdnZ2tvVXqNHQNdIztbQwsNDRNdAz1rHWNdQzsrQ0MNEx0DMx1bE2gAqjiqJo0awBAGUSD6K5ckndAAAAAElFTkSuQmCC",
      "text/plain": [
       "<rdkit.Chem.rdchem.Mol at 0x23e56c0c120>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Chem.MolFromSmiles(\"CCCC=O\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "Checking the output of ``set_attributes`` method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[array([[ 1.        ,  0.        ,  0.        ,  3.        , -0.06564544],\n",
       "        [ 1.        ,  0.        ,  0.        ,  2.        , -0.05903836],\n",
       "        [ 1.        ,  0.        ,  0.        ,  3.        , -0.06564544]],\n",
       "       dtype=float32),\n",
       " array([[ 1.        ,  0.        ,  0.        ,  3.        , -0.06319688],\n",
       "        [ 1.        ,  0.        ,  0.        ,  2.        , -0.03273489],\n",
       "        [ 1.        ,  0.        ,  0.        ,  2.        ,  0.04278928],\n",
       "        [ 0.        ,  0.        ,  1.        ,  1.        , -0.39638734]],\n",
       "       dtype=float32),\n",
       " array([[ 1.        ,  0.        ,  0.        ,  3.        , -0.06417979],\n",
       "        [ 1.        ,  0.        ,  0.        ,  2.        , -0.04375392],\n",
       "        [ 1.        ,  0.        ,  0.        ,  2.        , -0.00799057],\n",
       "        [ 0.        ,  1.        ,  0.        ,  2.        , -0.3304839 ]],\n",
       "       dtype=float32),\n",
       " array([[ 1.0000000e+00,  0.0000000e+00,  0.0000000e+00,  3.0000000e+00,\n",
       "         -6.4970836e-02],\n",
       "        [ 1.0000000e+00,  0.0000000e+00,  0.0000000e+00,  2.0000000e+00,\n",
       "         -4.9520925e-02],\n",
       "        [ 1.0000000e+00,  0.0000000e+00,  0.0000000e+00,  2.0000000e+00,\n",
       "          2.7463208e-03],\n",
       "        [ 1.0000000e+00,  0.0000000e+00,  0.0000000e+00,  1.0000000e+00,\n",
       "          1.1947978e-01],\n",
       "        [ 0.0000000e+00,  0.0000000e+00,  1.0000000e+00,  0.0000000e+00,\n",
       "         -3.0340311e-01]], dtype=float32),\n",
       " array([[ 0.        ,  1.        ,  0.        ,  2.        , -0.21275702],\n",
       "        [ 0.        ,  0.        ,  1.        ,  0.        , -0.2708128 ],\n",
       "        [ 1.        ,  0.        ,  0.        ,  2.        ,  0.2071647 ],\n",
       "        [ 0.        ,  0.        ,  0.        ,  0.        , -0.21796767]],\n",
       "       dtype=float32)]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dts.get(\"node_attributes\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[array([[1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.]], dtype=float32),\n",
       " array([[1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.]], dtype=float32),\n",
       " array([[1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.]], dtype=float32),\n",
       " array([[1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [2., 0.],\n",
       "        [2., 0.]], dtype=float32),\n",
       " array([[1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.]], dtype=float32)]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dts.get(\"edge_attributes\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[array([44.0626,  3.    ], dtype=float32),\n",
       " array([60.057514,  4.      ], dtype=float32),\n",
       " array([59.073498,  4.      ], dtype=float32),\n",
       " array([72.05752,  5.     ], dtype=float32),\n",
       " array([65.027695,  4.      ], dtype=float32)]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dts.get(\"graph_attributes\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[array([1, 0.1], dtype=object),\n",
       " array([2, 0.3], dtype=object),\n",
       " array([3, 0.2], dtype=object),\n",
       " array([4, 0.4], dtype=object),\n",
       " array([4, 1.4], dtype=object)]"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dts.get(\"graph_labels\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:kgcnn.data.ExampleMol:Pickle dataset...\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<MoleculeNetDataset [{'node_symbol': array(['C', 'C', 'C'], dtype='<U1'), 'node_number': array([6, 6, 6]), 'edge_indices': array([[0, 1],\n",
       "       [1, 0],\n",
       "       [1, 2],\n",
       "       [2, 1]], dtype=int64), 'edge_number': array([1, 1, 1, 1]), 'graph_size': array(3), 'node_coordinates': array([[ 1.2106, -0.1633, -0.3777],\n",
       "       [ 0.0398, -0.1833,  0.5905],\n",
       "       [-1.2407,  0.3016, -0.0679]]), 'graph_labels': array([1, 0.1], dtype=object), 'node_attributes': array([[ 1.        ,  0.        ,  0.        ,  3.        , -0.06564544],\n",
       "       [ 1.        ,  0.        ,  0.        ,  2.        , -0.05903836],\n",
       "       [ 1.        ,  0.        ,  0.        ,  3.        , -0.06564544]],\n",
       "      dtype=float32), 'edge_attributes': array([[1., 0.],\n",
       "       [1., 0.],\n",
       "       [1., 0.],\n",
       "       [1., 0.]], dtype=float32), 'graph_attributes': array([44.0626,  3.    ], dtype=float32), 'size': array(3)} ...]>"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dts.save()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:kgcnn.data.ExampleMol:Load pickled dataset...\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<MoleculeNetDataset [{'node_symbol': array(['C', 'C', 'C'], dtype='<U1'), 'node_number': array([6, 6, 6]), 'edge_indices': array([[0, 1],\n",
       "       [1, 0],\n",
       "       [1, 2],\n",
       "       [2, 1]], dtype=int64), 'edge_number': array([1, 1, 1, 1]), 'graph_size': array(3), 'node_coordinates': array([[ 1.2106, -0.1633, -0.3777],\n",
       "       [ 0.0398, -0.1833,  0.5905],\n",
       "       [-1.2407,  0.3016, -0.0679]]), 'graph_labels': array([1, 0.1], dtype=object), 'node_attributes': array([[ 1.        ,  0.        ,  0.        ,  3.        , -0.06564544],\n",
       "       [ 1.        ,  0.        ,  0.        ,  2.        , -0.05903836],\n",
       "       [ 1.        ,  0.        ,  0.        ,  3.        , -0.06564544]],\n",
       "      dtype=float32), 'edge_attributes': array([[1., 0.],\n",
       "       [1., 0.],\n",
       "       [1., 0.],\n",
       "       [1., 0.]], dtype=float32), 'graph_attributes': array([44.0626,  3.    ], dtype=float32), 'size': array(3)} ...]>"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dts.load()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'node_symbol': array(['C', 'C', 'C'], dtype='<U1'),\n",
       " 'node_number': array([6, 6, 6]),\n",
       " 'edge_indices': array([[0, 1],\n",
       "        [1, 0],\n",
       "        [1, 2],\n",
       "        [2, 1]], dtype=int64),\n",
       " 'edge_number': array([1, 1, 1, 1]),\n",
       " 'graph_size': array(3),\n",
       " 'node_coordinates': array([[ 1.2106, -0.1633, -0.3777],\n",
       "        [ 0.0398, -0.1833,  0.5905],\n",
       "        [-1.2407,  0.3016, -0.0679]]),\n",
       " 'graph_labels': array([1, 0.1], dtype=object),\n",
       " 'node_attributes': array([[ 1.        ,  0.        ,  0.        ,  3.        , -0.06564544],\n",
       "        [ 1.        ,  0.        ,  0.        ,  2.        , -0.05903836],\n",
       "        [ 1.        ,  0.        ,  0.        ,  3.        , -0.06564544]],\n",
       "       dtype=float32),\n",
       " 'edge_attributes': array([[1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.],\n",
       "        [1., 0.]], dtype=float32),\n",
       " 'graph_attributes': array([44.0626,  3.    ], dtype=float32),\n",
       " 'size': array(3)}"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dts[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}