example.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Google Colab setup - Uncomment to set up the environment in Google Colab\n",
    "# !git clone https://github.com/ml-jku/diverse-hits.git \n",
    "# !mv diverse-hits/* ./\n",
    "# !pip install -e ."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Example optimization\n",
    "This notebook outlines how:\n",
    "- the scoring function works\n",
    "- to create your custom scoring function\n",
    "- to adapt an optimizer to the scoring function.\n",
    "- the model is evaluated on the diverse optimization benchmark."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "from typing import List\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "from rdkit import RDLogger\n",
    "\n",
    "from divopt.evaluation.process_results import compute_diverse_solutions\n",
    "from divopt.scoring import BenchmarkScoringFunction\n",
    "\n",
    "RDLogger.DisableLog(\"rdApp.*\")\n",
    "plt.style.use(\"ggplot\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [],
   "source": [
    "# run the method until both budgets are exhausted. Default settings in the paper are 600s and 10k samples\n",
    "time_budget = 300\n",
    "sample_budget = 10000\n",
    "\n",
    "# set the memory parameters. Set both to None to disable memory\n",
    "memory_distance_threshold = 0.7\n",
    "memory_score_threshold = 0.5\n",
    "\n",
    "# some global parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [],
   "source": [
    "scoring_function_dir = \"data/scoring_functions/drd2\"\n",
    "scoring_function = BenchmarkScoringFunction(\n",
    "    scoring_function_dir=scoring_function_dir,\n",
    "    time_budget=time_budget,\n",
    "    sample_budget=sample_budget,\n",
    "    memory_distance_threshold=memory_distance_threshold,\n",
    "    memory_score_threshold=memory_score_threshold,\n",
    "    memory_known_active_init=False,\n",
    "    use_property_constraints=True,\n",
    "    n_jobs=8,\n",
    "    print_progress=True,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The scoring function can be called using a list of SMILES via the `score_list` method, \n",
    "which is compatible with the `guacamol' scoring function interface.\n",
    "\n",
    "The `__call__` method of the scoring function provides a wrapper around `score_list`.\n",
    "This method can also be used on a single SMILES string rather than a list.\n",
    "This provides compatibility with the oracle class used in `mol_opt`.\n",
    "\n",
    "The scoring function already take care of caching and checking the validity of the SMILES. Internally the memory/diversity filter is applied if the respective arguments `memory_score_threshold` and `memory_distance_threshold` are set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 109,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "with open(os.path.join(\"test/test_smiles.txt\"), \"r\") as f:\n",
    "    smiles_list = f.read().splitlines()[:20]\n",
    "\n",
    "scoring_function.start_timer_and_reset()\n",
    "scoring_function.print_progress = False\n",
    "\n",
    "scores0 = scoring_function.score_list(smiles_list)\n",
    "scores1 = scoring_function(smiles_list)\n",
    "scores2 = [scoring_function(s) for s in smiles_list]\n",
    "\n",
    "scores0 == scores1 == scores2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Optimization\n",
    "The scoring function interface assumes that the optimizer is based on an inner loop that jumps between a generation and scoring step. To get your own optimizer running on the benchmark, you need to implement a class with the following interface:\n",
    "\n",
    "```python\n",
    "class SomeOptimizer:\n",
    "    def __init__(self, arg1, ...):\n",
    "        self.arg1 = arg1\n",
    "        # ...\n",
    "\n",
    "    def generate_optimized_molecules(self, scoring_function):\n",
    "        while not finished:\n",
    "            # ...\n",
    "            scores = scoring_function.score_list(smiles_list)\n",
    "            self.optimization_step(scores)\n",
    "            # ...\n",
    "```\n",
    "\n",
    "In some cases it might be necessary to adapt the code of the optimizer. In some cases \n",
    "it suffices to add a wrapper to the scoring function. \n",
    "\n",
    "When this is done, the optimizer \n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from optimizers.guacamol_baselines.graph_ga.goal_directed_generation import GB_GA_Generator\n",
    "\n",
    "!wget -nc -q https://ndownloader.figshare.com/files/13612745 -O data/guacamol_v1_all.smiles \n",
    "optimizer = GB_GA_Generator(\n",
    "    smi_file=\"data/guacamol_v1_all.smiles\",\n",
    "    population_size=415,\n",
    "    offspring_size=232,\n",
    "    generations=1_000_000,\n",
    "    mutation_rate=0.05845893692037388,\n",
    "    n_jobs=8,\n",
    "    random_start=True,\n",
    "    patience=1000000000000000,\n",
    "    canonicalize=False,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loop 0000000 | Mean score: 0.017 | Max score: 0.370 | Memsize: 0003, Valid 1.000 | 676.729 mols/sec\n",
      "Loop 0000001 | Mean score: 0.016 | Max score: 0.370 | Memsize: 0003, Valid 1.000 | 4898.040 mols/sec\n",
      "Loop 0000002 | Mean score: 0.034 | Max score: 0.490 | Memsize: 0003, Valid 1.000 | 820.869 mols/sec\n",
      "Loop 0000003 | Mean score: 0.068 | Max score: 0.490 | Memsize: 0005, Valid 1.000 | 786.538 mols/sec\n",
      "Loop 0000004 | Mean score: 0.092 | Max score: 0.490 | Memsize: 0006, Valid 1.000 | 738.828 mols/sec\n",
      "Loop 0000005 | Mean score: 0.114 | Max score: 0.490 | Memsize: 0009, Valid 1.000 | 702.693 mols/sec\n",
      "Loop 0000006 | Mean score: 0.119 | Max score: 0.490 | Memsize: 0009, Valid 1.000 | 577.023 mols/sec\n",
      "Loop 0000007 | Mean score: 0.155 | Max score: 0.490 | Memsize: 0009, Valid 1.000 | 718.565 mols/sec\n",
      "Loop 0000008 | Mean score: 0.179 | Max score: 0.490 | Memsize: 0010, Valid 1.000 | 686.644 mols/sec\n",
      "Loop 0000009 | Mean score: 0.181 | Max score: 0.490 | Memsize: 0011, Valid 1.000 | 683.970 mols/sec\n",
      "Loop 0000010 | Mean score: 0.171 | Max score: 0.490 | Memsize: 0011, Valid 1.000 | 706.785 mols/sec\n",
      "Loop 0000011 | Mean score: 0.199 | Max score: 0.490 | Memsize: 0011, Valid 1.000 | 702.519 mols/sec\n",
      "Loop 0000012 | Mean score: 0.208 | Max score: 0.490 | Memsize: 0013, Valid 1.000 | 683.409 mols/sec\n",
      "Loop 0000013 | Mean score: 0.174 | Max score: 0.490 | Memsize: 0014, Valid 1.000 | 681.663 mols/sec\n",
      "Loop 0000014 | Mean score: 0.190 | Max score: 0.490 | Memsize: 0015, Valid 1.000 | 666.720 mols/sec\n",
      "Loop 0000015 | Mean score: 0.123 | Max score: 0.490 | Memsize: 0015, Valid 1.000 | 699.292 mols/sec\n",
      "Loop 0000016 | Mean score: 0.170 | Max score: 0.490 | Memsize: 0016, Valid 1.000 | 676.329 mols/sec\n",
      "Loop 0000017 | Mean score: 0.200 | Max score: 0.490 | Memsize: 0016, Valid 1.000 | 679.786 mols/sec\n",
      "Loop 0000018 | Mean score: 0.210 | Max score: 0.490 | Memsize: 0016, Valid 1.000 | 662.434 mols/sec\n",
      "Loop 0000019 | Mean score: 0.219 | Max score: 0.490 | Memsize: 0017, Valid 1.000 | 653.118 mols/sec\n",
      "Loop 0000020 | Mean score: 0.212 | Max score: 0.490 | Memsize: 0017, Valid 1.000 | 653.610 mols/sec\n",
      "Loop 0000021 | Mean score: 0.226 | Max score: 0.490 | Memsize: 0018, Valid 1.000 | 648.059 mols/sec\n",
      "Loop 0000022 | Mean score: 0.227 | Max score: 0.490 | Memsize: 0018, Valid 1.000 | 672.845 mols/sec\n",
      "Loop 0000023 | Mean score: 0.233 | Max score: 0.490 | Memsize: 0018, Valid 1.000 | 525.214 mols/sec\n",
      "Loop 0000024 | Mean score: 0.235 | Max score: 0.490 | Memsize: 0019, Valid 1.000 | 528.972 mols/sec\n",
      "Loop 0000025 | Mean score: 0.230 | Max score: 0.490 | Memsize: 0019, Valid 1.000 | 534.360 mols/sec\n",
      "Loop 0000026 | Mean score: 0.241 | Max score: 0.490 | Memsize: 0019, Valid 1.000 | 557.103 mols/sec\n",
      "Loop 0000027 | Mean score: 0.247 | Max score: 0.490 | Memsize: 0020, Valid 1.000 | 664.644 mols/sec\n",
      "Loop 0000028 | Mean score: 0.183 | Max score: 0.490 | Memsize: 0020, Valid 1.000 | 669.362 mols/sec\n",
      "Loop 0000029 | Mean score: 0.220 | Max score: 0.490 | Memsize: 0020, Valid 1.000 | 648.853 mols/sec\n",
      "Loop 0000030 | Mean score: 0.229 | Max score: 0.490 | Memsize: 0020, Valid 1.000 | 628.551 mols/sec\n",
      "Loop 0000031 | Mean score: 0.237 | Max score: 0.490 | Memsize: 0022, Valid 1.000 | 637.551 mols/sec\n",
      "Loop 0000032 | Mean score: 0.221 | Max score: 0.480 | Memsize: 0023, Valid 1.000 | 663.310 mols/sec\n",
      "Loop 0000033 | Mean score: 0.149 | Max score: 0.490 | Memsize: 0023, Valid 1.000 | 664.312 mols/sec\n",
      "Loop 0000034 | Mean score: 0.197 | Max score: 0.490 | Memsize: 0024, Valid 1.000 | 681.835 mols/sec\n",
      "Loop 0000035 | Mean score: 0.194 | Max score: 0.490 | Memsize: 0024, Valid 1.000 | 682.086 mols/sec\n",
      "Loop 0000036 | Mean score: 0.228 | Max score: 0.490 | Memsize: 0025, Valid 1.000 | 702.123 mols/sec\n",
      "Loop 0000037 | Mean score: 0.202 | Max score: 0.490 | Memsize: 0027, Valid 1.000 | 686.188 mols/sec\n",
      "Loop 0000038 | Mean score: 0.210 | Max score: 0.490 | Memsize: 0028, Valid 1.000 | 658.409 mols/sec\n",
      "Loop 0000039 | Mean score: 0.214 | Max score: 0.490 | Memsize: 0028, Valid 1.000 | 660.770 mols/sec\n",
      "Loop 0000040 | Mean score: 0.229 | Max score: 0.490 | Memsize: 0029, Valid 1.000 | 631.461 mols/sec\n",
      "Loop 0000041 | Mean score: 0.228 | Max score: 0.490 | Memsize: 0029, Valid 1.000 | 549.232 mols/sec\n",
      "Loop 0000042 | Mean score: 0.232 | Max score: 0.490 | Memsize: 0029, Valid 1.000 | 534.618 mols/sec\n",
      "Loop 0000043 | Mean score: 0.232 | Max score: 0.490 | Memsize: 0031, Valid 1.000 | 605.381 mols/sec\n",
      "Loop 0000044 | Mean score: 0.214 | Max score: 0.490 | Memsize: 0033, Valid 1.000 | 655.656 mols/sec\n",
      "Loop 0000045 | Mean score: 0.206 | Max score: 0.490 | Memsize: 0033, Valid 1.000 | 646.056 mols/sec\n",
      "Loop 0000046 | Mean score: 0.234 | Max score: 0.490 | Memsize: 0033, Valid 1.000 | 670.927 mols/sec\n",
      "Loop 0000047 | Mean score: 0.239 | Max score: 0.490 | Memsize: 0033, Valid 1.000 | 661.390 mols/sec\n",
      "Loop 0000048 | Mean score: 0.248 | Max score: 0.490 | Memsize: 0033, Valid 1.000 | 636.675 mols/sec\n",
      "Loop 0000049 | Mean score: 0.247 | Max score: 0.490 | Memsize: 0033, Valid 1.000 | 634.903 mols/sec\n",
      "Loop 0000050 | Mean score: 0.243 | Max score: 0.490 | Memsize: 0034, Valid 1.000 | 627.342 mols/sec\n",
      "Loop 0000051 | Mean score: 0.199 | Max score: 0.490 | Memsize: 0036, Valid 1.000 | 623.806 mols/sec\n",
      "Loop 0000052 | Mean score: 0.219 | Max score: 0.490 | Memsize: 0037, Valid 1.000 | 661.037 mols/sec\n",
      "Loop 0000053 | Mean score: 0.236 | Max score: 0.490 | Memsize: 0040, Valid 1.000 | 636.931 mols/sec\n",
      "Loop 0000054 | Mean score: 0.224 | Max score: 0.490 | Memsize: 0040, Valid 1.000 | 611.505 mols/sec\n",
      "Loop 0000055 | Mean score: 0.235 | Max score: 0.490 | Memsize: 0041, Valid 1.000 | 601.894 mols/sec\n",
      "Loop 0000056 | Mean score: 0.216 | Max score: 0.490 | Memsize: 0042, Valid 1.000 | 603.379 mols/sec\n",
      "Loop 0000057 | Mean score: 0.211 | Max score: 0.490 | Memsize: 0042, Valid 1.000 | 615.463 mols/sec\n",
      "Loop 0000058 | Mean score: 0.235 | Max score: 0.490 | Memsize: 0042, Valid 1.000 | 611.138 mols/sec\n",
      "Loop 0000059 | Mean score: 0.240 | Max score: 0.490 | Memsize: 0043, Valid 1.000 | 612.034 mols/sec\n",
      "Loop 0000060 | Mean score: 0.236 | Max score: 0.490 | Memsize: 0043, Valid 1.000 | 611.528 mols/sec\n",
      "Loop 0000061 | Mean score: 0.241 | Max score: 0.490 | Memsize: 0043, Valid 1.000 | 535.273 mols/sec\n",
      "Loop 0000062 | Mean score: 0.246 | Max score: 0.490 | Memsize: 0043, Valid 1.000 | 602.510 mols/sec\n",
      "Loop 0000063 | Mean score: 0.241 | Max score: 0.490 | Memsize: 0044, Valid 1.000 | 604.437 mols/sec\n",
      "Loop 0000064 | Mean score: 0.234 | Max score: 0.490 | Memsize: 0044, Valid 1.000 | 612.301 mols/sec\n",
      "Loop 0000065 | Mean score: 0.251 | Max score: 0.490 | Memsize: 0044, Valid 1.000 | 618.706 mols/sec\n",
      "Loop 0000066 | Mean score: 0.246 | Max score: 0.490 | Memsize: 0044, Valid 1.000 | 619.027 mols/sec\n",
      "Loop 0000067 | Mean score: 0.244 | Max score: 0.490 | Memsize: 0044, Valid 1.000 | 607.367 mols/sec\n",
      "Loop 0000068 | Mean score: 0.257 | Max score: 0.490 | Memsize: 0044, Valid 1.000 | 621.126 mols/sec\n",
      "Loop 0000069 | Mean score: 0.254 | Max score: 0.490 | Memsize: 0044, Valid 1.000 | 619.535 mols/sec\n",
      "Loop 0000070 | Mean score: 0.259 | Max score: 0.490 | Memsize: 0044, Valid 1.000 | 612.184 mols/sec\n",
      "Loop 0000071 | Mean score: 0.254 | Max score: 0.490 | Memsize: 0045, Valid 1.000 | 598.273 mols/sec\n",
      "Loop 0000072 | Mean score: 0.220 | Max score: 0.490 | Memsize: 0045, Valid 1.000 | 537.499 mols/sec\n",
      "Loop 0000073 | Mean score: 0.248 | Max score: 0.490 | Memsize: 0046, Valid 1.000 | 504.657 mols/sec\n",
      "Loop 0000074 | Mean score: 0.242 | Max score: 0.490 | Memsize: 0046, Valid 1.000 | 559.261 mols/sec\n",
      "Loop 0000075 | Mean score: 0.250 | Max score: 0.490 | Memsize: 0046, Valid 1.000 | 629.162 mols/sec\n",
      "Loop 0000076 | Mean score: 0.254 | Max score: 0.490 | Memsize: 0047, Valid 1.000 | 628.479 mols/sec\n",
      "Loop 0000077 | Mean score: 0.253 | Max score: 0.495 | Memsize: 0047, Valid 1.000 | 642.359 mols/sec\n",
      "Loop 0000078 | Mean score: 0.254 | Max score: 0.495 | Memsize: 0050, Valid 1.000 | 619.661 mols/sec\n",
      "Loop 0000079 | Mean score: 0.225 | Max score: 0.490 | Memsize: 0052, Valid 1.000 | 590.948 mols/sec\n",
      "Loop 0000080 | Mean score: 0.219 | Max score: 0.490 | Memsize: 0053, Valid 1.000 | 592.439 mols/sec\n",
      "Loop 0000081 | Mean score: 0.238 | Max score: 0.490 | Memsize: 0054, Valid 1.000 | 605.823 mols/sec\n",
      "Loop 0000082 | Mean score: 0.217 | Max score: 0.490 | Memsize: 0055, Valid 1.000 | 515.579 mols/sec\n",
      "Loop 0000083 | Mean score: 0.216 | Max score: 0.490 | Memsize: 0056, Valid 1.000 | 512.621 mols/sec\n",
      "Loop 0000084 | Mean score: 0.206 | Max score: 0.490 | Memsize: 0056, Valid 1.000 | 492.408 mols/sec\n",
      "Loop 0000085 | Mean score: 0.242 | Max score: 0.490 | Memsize: 0056, Valid 1.000 | 494.868 mols/sec\n",
      "Loop 0000086 | Mean score: 0.244 | Max score: 0.490 | Memsize: 0056, Valid 1.000 | 488.477 mols/sec\n",
      "Loop 0000087 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0057, Valid 1.000 | 560.106 mols/sec\n",
      "Loop 0000088 | Mean score: 0.247 | Max score: 0.490 | Memsize: 0057, Valid 1.000 | 600.070 mols/sec\n",
      "Loop 0000089 | Mean score: 0.255 | Max score: 0.490 | Memsize: 0057, Valid 1.000 | 530.942 mols/sec\n",
      "Loop 0000090 | Mean score: 0.255 | Max score: 0.490 | Memsize: 0058, Valid 1.000 | 493.905 mols/sec\n",
      "Loop 0000091 | Mean score: 0.236 | Max score: 0.480 | Memsize: 0058, Valid 1.000 | 592.635 mols/sec\n",
      "Loop 0000092 | Mean score: 0.257 | Max score: 0.490 | Memsize: 0058, Valid 1.000 | 589.984 mols/sec\n",
      "Loop 0000093 | Mean score: 0.250 | Max score: 0.490 | Memsize: 0059, Valid 1.000 | 543.771 mols/sec\n",
      "Loop 0000094 | Mean score: 0.198 | Max score: 0.490 | Memsize: 0059, Valid 1.000 | 487.113 mols/sec\n",
      "Loop 0000095 | Mean score: 0.237 | Max score: 0.490 | Memsize: 0059, Valid 1.000 | 605.765 mols/sec\n",
      "Loop 0000096 | Mean score: 0.249 | Max score: 0.490 | Memsize: 0059, Valid 1.000 | 542.897 mols/sec\n",
      "Loop 0000097 | Mean score: 0.246 | Max score: 0.490 | Memsize: 0059, Valid 1.000 | 621.596 mols/sec\n",
      "Loop 0000098 | Mean score: 0.243 | Max score: 0.490 | Memsize: 0060, Valid 1.000 | 526.383 mols/sec\n",
      "Loop 0000099 | Mean score: 0.236 | Max score: 0.490 | Memsize: 0061, Valid 1.000 | 558.978 mols/sec\n",
      "Loop 0000100 | Mean score: 0.248 | Max score: 0.490 | Memsize: 0061, Valid 1.000 | 574.497 mols/sec\n",
      "Loop 0000101 | Mean score: 0.243 | Max score: 0.490 | Memsize: 0063, Valid 1.000 | 636.672 mols/sec\n",
      "Loop 0000102 | Mean score: 0.176 | Max score: 0.490 | Memsize: 0064, Valid 1.000 | 634.029 mols/sec\n",
      "Loop 0000103 | Mean score: 0.213 | Max score: 0.490 | Memsize: 0064, Valid 1.000 | 644.236 mols/sec\n",
      "Loop 0000104 | Mean score: 0.230 | Max score: 0.490 | Memsize: 0064, Valid 1.000 | 625.942 mols/sec\n",
      "Loop 0000105 | Mean score: 0.240 | Max score: 0.490 | Memsize: 0064, Valid 1.000 | 610.874 mols/sec\n",
      "Loop 0000106 | Mean score: 0.240 | Max score: 0.490 | Memsize: 0065, Valid 1.000 | 620.656 mols/sec\n",
      "Loop 0000107 | Mean score: 0.205 | Max score: 0.480 | Memsize: 0065, Valid 1.000 | 656.766 mols/sec\n",
      "Loop 0000108 | Mean score: 0.238 | Max score: 0.480 | Memsize: 0065, Valid 1.000 | 586.917 mols/sec\n",
      "Loop 0000109 | Mean score: 0.239 | Max score: 0.480 | Memsize: 0065, Valid 1.000 | 639.229 mols/sec\n",
      "Loop 0000110 | Mean score: 0.247 | Max score: 0.490 | Memsize: 0066, Valid 1.000 | 593.722 mols/sec\n",
      "Loop 0000111 | Mean score: 0.240 | Max score: 0.490 | Memsize: 0067, Valid 1.000 | 611.700 mols/sec\n",
      "Loop 0000112 | Mean score: 0.244 | Max score: 0.490 | Memsize: 0067, Valid 1.000 | 599.252 mols/sec\n",
      "Loop 0000113 | Mean score: 0.249 | Max score: 0.490 | Memsize: 0067, Valid 1.000 | 612.544 mols/sec\n",
      "Loop 0000114 | Mean score: 0.248 | Max score: 0.490 | Memsize: 0067, Valid 1.000 | 619.028 mols/sec\n",
      "Loop 0000115 | Mean score: 0.241 | Max score: 0.490 | Memsize: 0068, Valid 1.000 | 611.321 mols/sec\n",
      "Loop 0000116 | Mean score: 0.214 | Max score: 0.480 | Memsize: 0068, Valid 1.000 | 531.883 mols/sec\n",
      "Loop 0000117 | Mean score: 0.239 | Max score: 0.490 | Memsize: 0069, Valid 1.000 | 565.602 mols/sec\n",
      "Loop 0000118 | Mean score: 0.226 | Max score: 0.490 | Memsize: 0069, Valid 1.000 | 613.874 mols/sec\n",
      "Loop 0000119 | Mean score: 0.242 | Max score: 0.490 | Memsize: 0071, Valid 1.000 | 542.722 mols/sec\n",
      "Loop 0000120 | Mean score: 0.232 | Max score: 0.490 | Memsize: 0072, Valid 1.000 | 541.449 mols/sec\n",
      "Loop 0000121 | Mean score: 0.232 | Max score: 0.490 | Memsize: 0072, Valid 1.000 | 544.116 mols/sec\n",
      "Loop 0000122 | Mean score: 0.240 | Max score: 0.490 | Memsize: 0074, Valid 1.000 | 622.109 mols/sec\n",
      "Loop 0000123 | Mean score: 0.207 | Max score: 0.490 | Memsize: 0074, Valid 1.000 | 630.003 mols/sec\n",
      "Loop 0000124 | Mean score: 0.236 | Max score: 0.490 | Memsize: 0074, Valid 1.000 | 589.048 mols/sec\n",
      "Loop 0000125 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0074, Valid 1.000 | 606.636 mols/sec\n",
      "Loop 0000126 | Mean score: 0.243 | Max score: 0.490 | Memsize: 0074, Valid 1.000 | 557.931 mols/sec\n",
      "Loop 0000127 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0075, Valid 1.000 | 506.104 mols/sec\n",
      "Loop 0000128 | Mean score: 0.243 | Max score: 0.490 | Memsize: 0075, Valid 1.000 | 506.483 mols/sec\n",
      "Loop 0000129 | Mean score: 0.246 | Max score: 0.490 | Memsize: 0075, Valid 1.000 | 501.771 mols/sec\n",
      "Loop 0000130 | Mean score: 0.257 | Max score: 0.490 | Memsize: 0075, Valid 1.000 | 425.516 mols/sec\n",
      "Loop 0000131 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0076, Valid 1.000 | 519.288 mols/sec\n",
      "Loop 0000132 | Mean score: 0.192 | Max score: 0.490 | Memsize: 0077, Valid 1.000 | 498.587 mols/sec\n",
      "Loop 0000133 | Mean score: 0.217 | Max score: 0.490 | Memsize: 0079, Valid 1.000 | 592.499 mols/sec\n",
      "Loop 0000134 | Mean score: 0.232 | Max score: 0.490 | Memsize: 0079, Valid 1.000 | 535.275 mols/sec\n",
      "Loop 0000135 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0079, Valid 1.000 | 577.176 mols/sec\n",
      "Loop 0000136 | Mean score: 0.246 | Max score: 0.490 | Memsize: 0080, Valid 1.000 | 570.936 mols/sec\n",
      "Loop 0000137 | Mean score: 0.247 | Max score: 0.490 | Memsize: 0081, Valid 1.000 | 595.300 mols/sec\n",
      "Loop 0000138 | Mean score: 0.242 | Max score: 0.490 | Memsize: 0082, Valid 1.000 | 570.703 mols/sec\n",
      "Loop 0000139 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0083, Valid 1.000 | 578.277 mols/sec\n",
      "Loop 0000140 | Mean score: 0.238 | Max score: 0.490 | Memsize: 0085, Valid 1.000 | 601.475 mols/sec\n",
      "Loop 0000141 | Mean score: 0.150 | Max score: 0.490 | Memsize: 0085, Valid 1.000 | 599.771 mols/sec\n",
      "Loop 0000142 | Mean score: 0.183 | Max score: 0.490 | Memsize: 0086, Valid 1.000 | 652.084 mols/sec\n",
      "Loop 0000143 | Mean score: 0.199 | Max score: 0.490 | Memsize: 0087, Valid 1.000 | 638.504 mols/sec\n",
      "Loop 0000144 | Mean score: 0.191 | Max score: 0.490 | Memsize: 0088, Valid 1.000 | 634.677 mols/sec\n",
      "Loop 0000145 | Mean score: 0.214 | Max score: 0.490 | Memsize: 0089, Valid 1.000 | 633.385 mols/sec\n",
      "Loop 0000146 | Mean score: 0.200 | Max score: 0.490 | Memsize: 0089, Valid 1.000 | 616.993 mols/sec\n",
      "Loop 0000147 | Mean score: 0.224 | Max score: 0.490 | Memsize: 0089, Valid 1.000 | 637.695 mols/sec\n",
      "Loop 0000148 | Mean score: 0.233 | Max score: 0.490 | Memsize: 0089, Valid 1.000 | 633.402 mols/sec\n",
      "Loop 0000149 | Mean score: 0.234 | Max score: 0.490 | Memsize: 0090, Valid 1.000 | 603.628 mols/sec\n",
      "Loop 0000150 | Mean score: 0.217 | Max score: 0.490 | Memsize: 0090, Valid 1.000 | 599.585 mols/sec\n",
      "Loop 0000151 | Mean score: 0.234 | Max score: 0.490 | Memsize: 0090, Valid 1.000 | 603.874 mols/sec\n",
      "Loop 0000152 | Mean score: 0.238 | Max score: 0.490 | Memsize: 0091, Valid 1.000 | 609.533 mols/sec\n",
      "Loop 0000153 | Mean score: 0.223 | Max score: 0.490 | Memsize: 0092, Valid 1.000 | 615.577 mols/sec\n",
      "Loop 0000154 | Mean score: 0.242 | Max score: 0.490 | Memsize: 0092, Valid 1.000 | 601.446 mols/sec\n",
      "Loop 0000155 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0093, Valid 1.000 | 600.291 mols/sec\n",
      "Loop 0000156 | Mean score: 0.244 | Max score: 0.490 | Memsize: 0093, Valid 1.000 | 622.292 mols/sec\n",
      "Loop 0000157 | Mean score: 0.244 | Max score: 0.490 | Memsize: 0093, Valid 1.000 | 605.584 mols/sec\n",
      "Loop 0000158 | Mean score: 0.244 | Max score: 0.490 | Memsize: 0094, Valid 1.000 | 607.672 mols/sec\n",
      "Loop 0000159 | Mean score: 0.231 | Max score: 0.490 | Memsize: 0097, Valid 1.000 | 579.667 mols/sec\n",
      "Loop 0000160 | Mean score: 0.170 | Max score: 0.490 | Memsize: 0098, Valid 1.000 | 609.189 mols/sec\n",
      "Loop 0000161 | Mean score: 0.194 | Max score: 0.490 | Memsize: 0101, Valid 1.000 | 607.681 mols/sec\n",
      "Loop 0000162 | Mean score: 0.182 | Max score: 0.490 | Memsize: 0101, Valid 1.000 | 589.602 mols/sec\n",
      "Loop 0000163 | Mean score: 0.223 | Max score: 0.490 | Memsize: 0101, Valid 1.000 | 573.314 mols/sec\n",
      "Loop 0000164 | Mean score: 0.236 | Max score: 0.490 | Memsize: 0102, Valid 1.000 | 593.684 mols/sec\n",
      "Loop 0000165 | Mean score: 0.242 | Max score: 0.490 | Memsize: 0102, Valid 1.000 | 534.799 mols/sec\n",
      "Loop 0000166 | Mean score: 0.250 | Max score: 0.490 | Memsize: 0103, Valid 1.000 | 497.994 mols/sec\n",
      "Loop 0000167 | Mean score: 0.201 | Max score: 0.490 | Memsize: 0105, Valid 1.000 | 530.024 mols/sec\n",
      "Loop 0000168 | Mean score: 0.232 | Max score: 0.490 | Memsize: 0106, Valid 1.000 | 530.050 mols/sec\n",
      "Loop 0000169 | Mean score: 0.178 | Max score: 0.490 | Memsize: 0106, Valid 1.000 | 528.317 mols/sec\n",
      "Loop 0000170 | Mean score: 0.217 | Max score: 0.490 | Memsize: 0106, Valid 1.000 | 572.661 mols/sec\n",
      "Loop 0000171 | Mean score: 0.240 | Max score: 0.490 | Memsize: 0106, Valid 1.000 | 537.546 mols/sec\n",
      "Loop 0000172 | Mean score: 0.237 | Max score: 0.490 | Memsize: 0107, Valid 1.000 | 553.798 mols/sec\n",
      "Loop 0000173 | Mean score: 0.212 | Max score: 0.490 | Memsize: 0107, Valid 1.000 | 555.404 mols/sec\n",
      "Loop 0000174 | Mean score: 0.244 | Max score: 0.490 | Memsize: 0107, Valid 1.000 | 526.444 mols/sec\n",
      "Loop 0000175 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0107, Valid 1.000 | 533.176 mols/sec\n",
      "Loop 0000176 | Mean score: 0.241 | Max score: 0.490 | Memsize: 0107, Valid 1.000 | 532.647 mols/sec\n",
      "Loop 0000177 | Mean score: 0.244 | Max score: 0.490 | Memsize: 0108, Valid 1.000 | 561.383 mols/sec\n",
      "Loop 0000178 | Mean score: 0.225 | Max score: 0.490 | Memsize: 0109, Valid 1.000 | 548.749 mols/sec\n",
      "Loop 0000179 | Mean score: 0.233 | Max score: 0.490 | Memsize: 0109, Valid 1.000 | 547.155 mols/sec\n",
      "Loop 0000180 | Mean score: 0.248 | Max score: 0.490 | Memsize: 0110, Valid 1.000 | 524.749 mols/sec\n",
      "Loop 0000181 | Mean score: 0.240 | Max score: 0.490 | Memsize: 0110, Valid 1.000 | 455.948 mols/sec\n",
      "Loop 0000182 | Mean score: 0.248 | Max score: 0.490 | Memsize: 0110, Valid 1.000 | 542.915 mols/sec\n",
      "Loop 0000183 | Mean score: 0.247 | Max score: 0.490 | Memsize: 0110, Valid 1.000 | 526.653 mols/sec\n",
      "Loop 0000184 | Mean score: 0.251 | Max score: 0.490 | Memsize: 0110, Valid 1.000 | 520.593 mols/sec\n",
      "Loop 0000185 | Mean score: 0.253 | Max score: 0.490 | Memsize: 0111, Valid 1.000 | 553.987 mols/sec\n",
      "Loop 0000186 | Mean score: 0.213 | Max score: 0.490 | Memsize: 0111, Valid 1.000 | 565.858 mols/sec\n",
      "Loop 0000187 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0112, Valid 1.000 | 553.412 mols/sec\n",
      "Loop 0000188 | Mean score: 0.244 | Max score: 0.490 | Memsize: 0112, Valid 1.000 | 566.895 mols/sec\n",
      "Loop 0000189 | Mean score: 0.249 | Max score: 0.490 | Memsize: 0112, Valid 1.000 | 534.870 mols/sec\n",
      "Loop 0000190 | Mean score: 0.251 | Max score: 0.490 | Memsize: 0112, Valid 1.000 | 545.897 mols/sec\n",
      "Loop 0000191 | Mean score: 0.259 | Max score: 0.490 | Memsize: 0113, Valid 1.000 | 533.887 mols/sec\n",
      "Loop 0000192 | Mean score: 0.253 | Max score: 0.490 | Memsize: 0114, Valid 1.000 | 544.150 mols/sec\n",
      "Loop 0000193 | Mean score: 0.234 | Max score: 0.490 | Memsize: 0116, Valid 1.000 | 546.166 mols/sec\n",
      "Loop 0000194 | Mean score: 0.182 | Max score: 0.490 | Memsize: 0116, Valid 1.000 | 544.036 mols/sec\n",
      "Loop 0000195 | Mean score: 0.224 | Max score: 0.490 | Memsize: 0117, Valid 1.000 | 545.895 mols/sec\n",
      "Loop 0000196 | Mean score: 0.228 | Max score: 0.490 | Memsize: 0117, Valid 1.000 | 546.254 mols/sec\n",
      "Loop 0000197 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0117, Valid 1.000 | 530.415 mols/sec\n",
      "Loop 0000198 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0117, Valid 1.000 | 525.858 mols/sec\n",
      "Loop 0000199 | Mean score: 0.245 | Max score: 0.490 | Memsize: 0118, Valid 1.000 | 507.074 mols/sec\n",
      "Loop 0000200 | Mean score: 0.156 | Max score: 0.490 | Memsize: 0118, Valid 1.000 | 503.366 mols/sec\n",
      "Loop 0000201 | Mean score: 0.191 | Max score: 0.490 | Memsize: 0118, Valid 1.000 | 463.343 mols/sec\n",
      "Loop 0000202 | Mean score: 0.221 | Max score: 0.490 | Memsize: 0118, Valid 1.000 | 558.378 mols/sec\n",
      "Loop 0000203 | Mean score: 0.235 | Max score: 0.490 | Memsize: 0119, Valid 1.000 | 559.313 mols/sec\n",
      "Loop 0000204 | Mean score: 0.215 | Max score: 0.490 | Memsize: 0119, Valid 1.000 | 543.795 mols/sec\n",
      "Loop 0000205 | Mean score: 0.241 | Max score: 0.490 | Memsize: 0119, Valid 1.000 | 545.025 mols/sec\n",
      "Loop 0000206 | Mean score: 0.237 | Max score: 0.490 | Memsize: 0121, Valid 1.000 | 540.964 mols/sec\n",
      "Loop 0000207 | Mean score: 0.198 | Max score: 0.490 | Memsize: 0122, Valid 1.000 | 542.350 mols/sec\n",
      "Loop 0000208 | Mean score: 0.152 | Max score: 0.490 | Memsize: 0122, Valid 1.000 | 549.355 mols/sec\n",
      "Loop 0000209 | Mean score: 0.190 | Max score: 0.490 | Memsize: 0122, Valid 1.000 | 547.429 mols/sec\n",
      "Loop 0000210 | Mean score: 0.218 | Max score: 0.490 | Memsize: 0122, Valid 1.000 | 551.845 mols/sec\n",
      "Loop 0000211 | Mean score: 0.217 | Max score: 0.490 | Memsize: 0123, Valid 1.000 | 546.700 mols/sec\n",
      "Loop 0000212 | Mean score: 0.210 | Max score: 0.490 | Memsize: 0123, Valid 1.000 | 556.112 mols/sec\n",
      "Loop 0000213 | Mean score: 0.225 | Max score: 0.490 | Memsize: 0123, Valid 1.000 | 516.065 mols/sec\n",
      "Loop 0000214 | Mean score: 0.230 | Max score: 0.490 | Memsize: 0123, Valid 1.000 | 480.724 mols/sec\n",
      "Compute budget exhausted\n"
     ]
    }
   ],
   "source": [
    "# this starts the timer and resets the cache.\n",
    "scoring_function.print_progress = True\n",
    "scoring_function.start_timer_and_reset()\n",
    "try:\n",
    "    optimizer.generate_optimized_molecules(scoring_function)\n",
    "except TimeoutError:\n",
    "    print(\"Compute budget exhausted\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CanSmiles</th>\n",
       "      <th>Score</th>\n",
       "      <th>Total time [s]</th>\n",
       "      <th>Scoring time [s]</th>\n",
       "      <th>Generation time [s]</th>\n",
       "      <th>Memory time [s]</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>175847</th>\n",
       "      <td>CCc1ccc2c(c1)CCC(CCN1C3CCC1CC(C=CCNCCCCCNCC=CC...</td>\n",
       "      <td>0.33</td>\n",
       "      <td>299.854617</td>\n",
       "      <td>89.381597</td>\n",
       "      <td>199.132819</td>\n",
       "      <td>11.340201</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>175848</th>\n",
       "      <td>N#CC1=CC=C(CC2=CCC(CCCCO)CC2)CC1</td>\n",
       "      <td>0.06</td>\n",
       "      <td>299.856697</td>\n",
       "      <td>89.382160</td>\n",
       "      <td>199.134187</td>\n",
       "      <td>11.340350</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>175849</th>\n",
       "      <td>N#CNCc1ccc2c(c1)CCC(C1=C(CCC3C=CC=CC3)CN(CNC=C...</td>\n",
       "      <td>0.28</td>\n",
       "      <td>299.858778</td>\n",
       "      <td>89.382723</td>\n",
       "      <td>199.135555</td>\n",
       "      <td>11.340500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>175850</th>\n",
       "      <td>OC1CCN(CC2=CCC(CC3CCC(CCN4CC=C(N5CCN(F)CC5)CC4...</td>\n",
       "      <td>0.00</td>\n",
       "      <td>299.860858</td>\n",
       "      <td>89.383286</td>\n",
       "      <td>199.136923</td>\n",
       "      <td>11.340649</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>175851</th>\n",
       "      <td>N#CNCCCNCC=CCl</td>\n",
       "      <td>0.01</td>\n",
       "      <td>299.862938</td>\n",
       "      <td>89.383849</td>\n",
       "      <td>199.138291</td>\n",
       "      <td>11.340798</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                CanSmiles  Score  \\\n",
       "175847  CCc1ccc2c(c1)CCC(CCN1C3CCC1CC(C=CCNCCCCCNCC=CC...   0.33   \n",
       "175848                   N#CC1=CC=C(CC2=CCC(CCCCO)CC2)CC1   0.06   \n",
       "175849  N#CNCc1ccc2c(c1)CCC(C1=C(CCC3C=CC=CC3)CN(CNC=C...   0.28   \n",
       "175850  OC1CCN(CC2=CCC(CC3CCC(CCN4CC=C(N5CCN(F)CC5)CC4...   0.00   \n",
       "175851                                     N#CNCCCNCC=CCl   0.01   \n",
       "\n",
       "        Total time [s]  Scoring time [s]  Generation time [s]  Memory time [s]  \n",
       "175847      299.854617         89.381597           199.132819        11.340201  \n",
       "175848      299.856697         89.382160           199.134187        11.340350  \n",
       "175849      299.858778         89.382723           199.135555        11.340500  \n",
       "175850      299.860858         89.383286           199.136923        11.340649  \n",
       "175851      299.862938         89.383849           199.138291        11.340798  "
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Now we can get the history of scored molecules from the scoring function\n",
    "df = scoring_function.get_history()\n",
    "df.tail()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluating the optimizer\n",
    "We can now evaluate the optimizer in a few steps to arrive at the number of unique and diverse hits.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use these values for evaluation. \n",
    "score_threshold = 0.5\n",
    "distance_threshold = 0.7\n",
    "\n",
    "# reduce the dataframe to unique molecules and add a column for how many molecules were scored so far\n",
    "df_unique = df.drop_duplicates(subset=\"CanSmiles\").copy()\n",
    "df_unique[\"#Scored molecules\"] = np.arange(len(df_unique))\n",
    "\n",
    "# only keep molecules that pass the score threshold\n",
    "df_unique_passing_score = df_unique[df_unique[\"Score\"] > score_threshold]\n",
    "\n",
    "# compute how many diverse molecules we have according to the #Circles metric.\n",
    "smiles_list: List[str] = df_unique_passing_score[\"CanSmiles\"].to_list()\n",
    "idx_divers = compute_diverse_solutions(smiles_list, distance_threshold=distance_threshold, algorithm=\"maxmin\")\n",
    "\n",
    "df_diverse = df_unique_passing_score.iloc[idx_divers]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot the results\n",
    "Finally, we can plot the results to see how the optimization went.\n",
    "Here we plot the number of unique and diverse hits as a function of the compute budget.\n",
    "The numbers given in the paper are the number of diverse hits once the respective compute budget is reached."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of diverse hits at sample budget: 13\n",
      "Number of diverse hits at time budget: 85\n"
     ]
    }
   ],
   "source": [
    "divhits_sample_budget = np.searchsorted(df_diverse[\"#Scored molecules\"], sample_budget)\n",
    "divhits_time_budget = np.searchsorted(df_diverse[\"Total time [s]\"], time_budget)\n",
    "\n",
    "print(f\"Number of diverse hits at sample budget: {divhits_sample_budget}\")\n",
    "print(f\"Number of diverse hits at time budget: {divhits_time_budget}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compute_measures = [\"Total time [s]\", \"#Scored molecules\"]\n",
    "compute_budgets = dict(zip(compute_measures, [time_budget, sample_budget]))\n",
    "divhits = dict(zip(compute_measures, [divhits_time_budget, divhits_sample_budget]))\n",
    "\n",
    "plot_dfs = {\n",
    "    \"Unique\": df_unique_passing_score,\n",
    "    f\"Diverse@{distance_threshold}\": df_diverse,\n",
    "}\n",
    "\n",
    "fig, axes = plt.subplots(1, ncols=len(compute_measures), figsize=(10, 4), sharey=True)\n",
    "\n",
    "for i, measure in enumerate(compute_measures):\n",
    "    for name, plot_df in plot_dfs.items():\n",
    "        xs = plot_df[measure]\n",
    "        ys = np.arange(len(plot_df))\n",
    "        axes[i].plot(xs.values, ys, label=name)\n",
    "\n",
    "    axes[i].axvline(compute_budgets[measure], color=\"grey\", linestyle=\"--\", lw=1)\n",
    "    axes[i].axhline(divhits[measure], color=\"grey\", linestyle=\"--\", lw=1)\n",
    "    axes[i].legend()\n",
    "    axes[i].set_xlabel(measure)\n",
    "axes[0].set_ylabel(\"#Hits\")\n",
    "plt.yscale(\"log\")\n",
    "plt.tight_layout()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base",
   "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": "3.11.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}