examples/protein_optimization/GB1/protein_optimization_GB1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "2a8260c1-6bd1-40bc-b1b4-fec70c75f12e",
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "d4a7a563-f646-411c-a4c2-f4609e9f9a58",
   "metadata": {},
   "outputs": [],
   "source": [
    "import yaml\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import torch\n",
    "from gpytorch.kernels import RBFKernel\n",
    "from gpytorch.priors import NormalPrior\n",
    "\n",
    "from mobius import Planner, SequenceGA\n",
    "from mobius import ProteinEmbedding\n",
    "from mobius import InverseFolding\n",
    "from mobius import GPLLModel, ExpectedImprovement\n",
    "from mobius.utils import generate_biopolymer_design_protocol_from_probabilities\n",
    "from mobius import homolog_scanning\n",
    "from mobius import convert_FASTA_to_HELM, convert_HELM_to_FASTA"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "522fb502-7d87-45c7-a561-7ec59107a6b2",
   "metadata": {},
   "source": [
    "## GB1 dataset\n",
    "\n",
    "Citation: Adaptation in protein fitness landscapes is facilitated by indirect paths; Wu et al.; 2016; http://dx.doi.org/10.7554/eLife.16965.001"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b2266942-7242-4bb9-8af0-b98b521dfa9a",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/scicore/home/schwede/eberha0001/Applications/mambaforge/envs/mobius-ubuntu/lib/python3.11/site-packages/openpyxl/worksheet/_read_only.py:85: UserWarning: Unknown extension is not supported and will be removed\n",
      "  for idx, row in parser.parse():\n",
      "/scicore/home/schwede/eberha0001/Applications/mambaforge/envs/mobius-ubuntu/lib/python3.11/site-packages/openpyxl/worksheet/_read_only.py:85: UserWarning: Unknown extension is not supported and will be removed\n",
      "  for idx, row in parser.parse():\n"
     ]
    }
   ],
   "source": [
    "df_exp = pd.read_excel('elife-16965-supp1-v4.xlsx')\n",
    "df_imputated = pd.read_excel('elife-16965-supp2-v4.xlsx')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "cfb3ebae-599e-490f-bf63-d64625626d7e",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_imputated = df_imputated.rename(columns={\"Imputed fitness\": \"Fitness\"})\n",
    "df = pd.concat([df_exp[['Variants', 'Fitness']], df_imputated])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "693fb597-bd00-47b5-b509-525fff40e3d1",
   "metadata": {},
   "outputs": [],
   "source": [
    "sequences = []\n",
    "wt_sequence = list('MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGVDGEWTYDDATKTFTVTE')\n",
    "\n",
    "for variant in df['Variants'].values:\n",
    "    wt_sequence[38] = variant[0]\n",
    "    wt_sequence[39] = variant[1]\n",
    "    wt_sequence[40] = variant[2]\n",
    "    wt_sequence[53] = variant[3]\n",
    "\n",
    "    sequences.append(''.join(wt_sequence))\n",
    "\n",
    "df['Sequences'] = sequences\n",
    "df.to_csv('GB1_dataset.csv', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bb93cdcd-2e57-463d-a47b-dbfc3cb6e9ed",
   "metadata": {},
   "source": [
    "## Initialize GB1 oracle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "2406fa99-156c-4ed8-b619-8b9a2985ff8f",
   "metadata": {},
   "outputs": [],
   "source": [
    "class GB1Scorer:\n",
    "    \n",
    "    def __init__(self, sequences, values):\n",
    "        if len(sequences) != len(values):\n",
    "            raise ValueError(f'Different numbers of sequences and values ({len(sequences)} != {len(values)})')\n",
    "\n",
    "        self._data = {s: v for s, v in zip(sequences, values)}\n",
    "\n",
    "    def score(self, sequences):\n",
    "        if not isinstance(sequences, (list, tuple, np.ndarray)):\n",
    "            sequences = [sequences]\n",
    "        \n",
    "        return np.array([self._data[s] for s in sequences])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b031b180-d8ba-4d74-bc91-9e6e7dd5a95c",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.])"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('GB1_dataset.csv')\n",
    "sequences = df['Sequences'].values\n",
    "fitness = df['Fitness'].values\n",
    "\n",
    "gb1 = GB1Scorer(sequences, fitness)\n",
    "gb1.score('MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGVDGEWTYDDATKTFTVTE')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ec96157c-d5c4-4ee7-8ece-5b32be8a1f4d",
   "metadata": {},
   "source": [
    "## Some plots..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "7bb72b46-2430-4de6-b906-5b820c00c1d3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0\n",
      "9.9130883273888\n"
     ]
    }
   ],
   "source": [
    "print(np.min(fitness))\n",
    "print(np.max(fitness))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "0010f89b-0dc7-4c95-9bfd-feb0b68394d1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 2000x750 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(20, 7.5))\n",
    "ax.hist(fitness, bins=np.linspace(0, 10, 101), log=True)\n",
    "#ax.set_ylim([-0.1, 500])\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0c8521fa-9945-4be8-aeb9-5b29fa51825b",
   "metadata": {},
   "source": [
    "## Run Bayesian Optimization"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "52a4d06c-d185-4133-b512-36319c981ca6",
   "metadata": {},
   "source": [
    "### With ESM-1b only"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "fff1b20e-df41-428c-8120-7d4b8bd5fc6b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load protein language model\n",
    "plm = ProteinEmbedding(pretrained_model_name='esm1b_t33_650M_UR50S', embedding_type='avg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "20825f8c-6b9f-40d8-b889-62f5313d7196",
   "metadata": {},
   "outputs": [],
   "source": [
    "yaml_content = \"\"\"\n",
    "    design:\n",
    "      monomers:\n",
    "        default: [A, C, D, E, F, G, H, I, K, L, M, N, P, Q, R, S, T, V, W, Y]\n",
    "      biopolymers:\n",
    "        - name: GB1\n",
    "          starting_residue: 1\n",
    "          length: 56\n",
    "          positions:\n",
    "            39: default\n",
    "            40: default\n",
    "            41: default\n",
    "            54: default\n",
    "    \"\"\"\n",
    "    \n",
    "with open('design_protocol_gb1.yaml', 'w') as f:\n",
    "    f.write(yaml_content)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "bb931dd9-07be-4b95-9d32-5bb393599258",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load surrogate model\n",
    "gpmodel = GPLLModel(kernel=RBFKernel(), pretrained_model=plm, noise_prior=NormalPrior(0, 1))\n",
    "ei = ExpectedImprovement(gpmodel, maximize=True)\n",
    "optimizer = SequenceGA(algorithm='GA', period=15, design_protocol_filename='design_protocol_gb1.yaml')\n",
    "ps = Planner(ei, optimizer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "11786c13-28da-4626-9534-7f59d5483ad8",
   "metadata": {},
   "outputs": [],
   "source": [
    "lead_protein = convert_FASTA_to_HELM('MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGVDGEWTYDDATKTFTVTE')[0]\n",
    "\n",
    "seed_library = [lead_protein]\n",
    "positions = {'PEPTIDE1': [39, 40, 41, 54]}\n",
    "\n",
    "for seq in homolog_scanning(lead_protein, positions=positions):\n",
    "    seed_library.append(seq)\n",
    "\n",
    "    if len(seed_library) >= 96:\n",
    "        print('Reach max. number of protein allowed.')\n",
    "        break\n",
    "\n",
    "seed_library = convert_HELM_to_FASTA(seed_library)\n",
    "scores_seed_library = gb1.score(seed_library)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "2ecce926-1f13-4d5f-8fa6-037932480e9b",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Fitting GPLLM model (cuda): 14step [00:15,  1.13s/step, loss=1.02]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==========================================================================================================\n",
      "n_gen  |  n_eval  |        f_avg         |        f_min         |              best sequence              \n",
      "==========================================================================================================\n",
      "     1 |        0 |  7.15435665664993881 |  3.89803670068332009 | MQYKLILNGKTLKGETTT....VWGEWTYDDATKTFTVTE\n",
      "     2 |      500 | -5.4890792508424E-04 | -4.7858192303212E-02 | MQYKLILNGKTLKGETTT....IYGEWTYDDATKTFTVTE\n",
      "     3 |     1000 | -9.8040651379193E-04 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     4 |     1500 | -1.2202246562406E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     5 |     2000 | -1.4845849976290E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     6 |     2500 | -1.7738633970282E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     7 |     3000 | -1.9960620675125E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     8 |     3500 | -2.1202157423563E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     9 |     4000 | -2.2032741551248E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    10 |     4500 | -2.2523480770819E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    11 |     5000 | -2.2877123313522E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    12 |     5500 | -2.3208215090008E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    13 |     6000 | -2.3298231532890E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    14 |     6500 | -2.3351096798517E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    15 |     7000 | -2.3499287532160E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    16 |     7500 | -2.3565248359374E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    17 |     8000 | -2.3584165072531E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    18 |     8500 | -2.3606598047801E-03 | -6.3728497800166E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "Best GB1 found so far: MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGIWGEWTYDDATKTFTFTE / 5.772 (min: 0.000, mean: 1.619, max: 5.772)\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Fitting GPLLM model (cuda): 11step [00:54,  4.93s/step, loss=1.26]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==========================================================================================================\n",
      "n_gen  |  n_eval  |        f_avg         |        f_min         |              best sequence              \n",
      "==========================================================================================================\n",
      "     1 |        0 |  1.0358340832886E+01 |  5.77203220933999983 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     2 |      500 | -4.5434898414752E-05 | -8.0754850842790E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     3 |     1000 | -8.1294941781521E-05 | -8.0754850842790E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     4 |     1500 | -1.5636922007626E-04 | -8.0754850842790E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     5 |     2000 | -2.5661998078577E-04 | -8.0754850842790E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     6 |     2500 | -3.3898723707837E-04 | -8.0754850842790E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     7 |     3000 | -4.1032354786453E-04 | -8.0754850842790E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     8 |     3500 | -4.6445791492326E-04 | -8.0754850842790E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     9 |     4000 | -5.0173388058402E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    10 |     4500 | -5.3027527134624E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    11 |     5000 | -5.5878603492055E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    12 |     5500 | -5.7867444672905E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    13 |     6000 | -5.8800071784080E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    14 |     6500 | -5.9275511679469E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    15 |     7000 | -5.9650593406491E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    16 |     7500 | -5.9892503042840E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    17 |     8000 | -5.9964637474387E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    18 |     8500 | -6.0017421205008E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    19 |     9000 | -6.0042075113588E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    20 |     9500 | -6.0086967990170E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    21 |    10000 | -6.0107943185918E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    22 |    10500 | -6.0120111137417E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    23 |    11000 | -6.0159084850051E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "    24 |    11500 | -6.0164589571289E-04 | -8.0957325514768E-03 | MQYKLILNGKTLKGETTT....FFGEWTYDDATKTFTWTE\n",
      "Best GB1 found so far: MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGIWGEWTYDDATKTFTFTE / 5.772 (min: 0.000, mean: 1.435, max: 5.772)\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Fitting GPLLM model (cuda): 16step [00:49,  3.08s/step, loss=1.2]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==========================================================================================================\n",
      "n_gen  |  n_eval  |        f_avg         |        f_min         |              best sequence              \n",
      "==========================================================================================================\n",
      "     1 |        0 |  1.0269218537592E+01 |  5.77203220933999983 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     2 |      500 | -2.1548256152679E-06 | -8.9629384173074E-05 | MQYKLILNGKTLKGETTT....LWGEWTYDDATKTFTGTE\n",
      "     3 |     1000 | -9.8634805228739E-06 | -5.7750539360324E-04 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     4 |     1500 | -2.6077173119337E-05 | -7.1319230768473E-04 | MQYKLILNGKTLKGETTT....VYCEWTYDDATKTFTGTE\n",
      "     5 |     2000 | -4.7637009444651E-05 | -7.1319230768473E-04 | MQYKLILNGKTLKGETTT....VYCEWTYDDATKTFTGTE\n",
      "     6 |     2500 | -7.9541313177828E-05 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "     7 |     3000 | -1.0751927049454E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "     8 |     3500 | -1.2944904269400E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "     9 |     4000 | -1.3986974683751E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    10 |     4500 | -1.4752156036144E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    11 |     5000 | -1.4942598704276E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    12 |     5500 | -1.5166660484291E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    13 |     6000 | -1.5307210541839E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    14 |     6500 | -1.5394505116821E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    15 |     7000 | -1.5482581833534E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    16 |     7500 | -1.5503907194629E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    17 |     8000 | -1.5512051494968E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    18 |     8500 | -1.5536434056384E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    19 |     9000 | -1.5539454084184E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    20 |     9500 | -1.5579118444078E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "    21 |    10000 | -1.5585437431376E-04 | -1.3125923215387E-03 | MQYKLILNGKTLKGETTT....LYCEWTYDDATKTFTGTE\n",
      "Best GB1 found so far: MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGIWGEWTYDDATKTFTFTE / 5.772 (min: 0.016, mean: 1.888, max: 5.772)\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Fitting GPLLM model (cuda): 17step [01:58,  6.94s/step, loss=1.23]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==========================================================================================================\n",
      "n_gen  |  n_eval  |        f_avg         |        f_min         |              best sequence              \n",
      "==========================================================================================================\n",
      "     1 |        0 |  1.0107880084476E+01 |  5.77203220933999983 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     2 |      500 | -4.4082553555864E-05 | -2.8222104587430E-03 | MQYKLILNGKTLKGETTT....YWFEWTYDDATKTFTGTE\n",
      "     3 |     1000 | -1.1531358567019E-04 | -1.7885506554139E-02 | MQYKLILNGKTLKGETTT....LWCEWTYDDATKTFTGTE\n",
      "     4 |     1500 | -2.9281294896485E-04 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "     5 |     2000 | -4.2390882097764E-04 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "     6 |     2500 | -5.7242254640993E-04 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "     7 |     3000 | -7.1565917589830E-04 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "     8 |     3500 | -8.3004664443381E-04 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "     9 |     4000 | -8.9135171823420E-04 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "    10 |     4500 | -9.2758617118871E-04 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "    11 |     5000 | -9.5966070647892E-04 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "    12 |     5500 | -9.7453780320128E-04 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "    13 |     6000 | -9.8748967770441E-04 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "    14 |     6500 | -9.9857595717732E-04 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "    15 |     7000 | -1.0034769622931E-03 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "    16 |     7500 | -1.0104934220874E-03 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "    17 |     8000 | -1.0167160431895E-03 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "    18 |     8500 | -1.0209093715634E-03 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "    19 |     9000 | -1.0271804984304E-03 | -2.7174249713992E-02 | MQYKLILNGKTLKGETTT....WWFEWTYDDATKTFTGTE\n",
      "Best GB1 found so far: MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGFWLEWTYDDATKTFTGTE / 7.313 (min: 0.000, mean: 1.093, max: 7.313)\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Fitting GPLLM model (cuda): 15step [02:43, 10.91s/step, loss=1.24]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==========================================================================================================\n",
      "n_gen  |  n_eval  |        f_avg         |        f_min         |              best sequence              \n",
      "==========================================================================================================\n",
      "     1 |        0 |  1.3260580700681E+01 |  7.31265639681999957 | MQYKLILNGKTLKGETTT....FWLEWTYDDATKTFTGTE\n",
      "     2 |      500 | -1.7630015076716E-06 | -4.9342076800113E-04 | MQYKLILNGKTLKGETTT....WKFEWTYDDATKTFTGTE\n",
      "     3 |     1000 | -1.3053133818884E-05 | -4.8662886842484E-03 | MQYKLILNGKTLKGETTT....FWLEWTYDDATKTFTGTE\n",
      "     4 |     1500 | -4.8513001363540E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "     5 |     2000 | -5.0579937236386E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "     6 |     2500 | -5.4409363830808E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "     7 |     3000 | -5.5089500578718E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "     8 |     3500 | -5.5290096238358E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "     9 |     4000 | -5.5341939276409E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "    10 |     4500 | -5.5571767215083E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "    11 |     5000 | -5.5711457911229E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "    12 |     5500 | -5.5733438149093E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "    13 |     6000 | -5.5761605328246E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "    14 |     6500 | -5.5812591752985E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "    15 |     7000 | -5.5853639467400E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "    16 |     7500 | -5.5865574672453E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "    17 |     8000 | -5.5881377899491E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "    18 |     8500 | -5.5894880923669E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "    19 |     9000 | -5.5921706805379E-05 | -1.6170238769609E-02 | MQYKLILNGKTLKGETTT....WWLEWTYDDATKTFTGTE\n",
      "Best GB1 found so far: MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGFWLEWTYDDATKTFTGTE / 7.313 (min: 0.010, mean: 2.226, max: 7.313)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "sequences = seed_library.copy()\n",
    "scores = scores_seed_library.copy()\n",
    "        \n",
    "# Store data for later analysis\n",
    "data = [(0, p, s) for p, s in zip(sequences, scores)]\n",
    "        \n",
    "for i in range(5):\n",
    "    suggested_sequences, _ = ps.recommand(sequences, scores.reshape(-1, 1), batch_size=96)\n",
    "    scores_suggested_sequences = gb1.score(suggested_sequences)\n",
    "        \n",
    "    sequences = np.concatenate([sequences, suggested_sequences])\n",
    "    scores = np.concatenate((scores, scores_suggested_sequences), axis=0)\n",
    "    data.extend([(i + 1, p, s) for p, s in zip(suggested_sequences, scores_suggested_sequences)])\n",
    "        \n",
    "    best_seq = sequences[np.argmax(scores)]\n",
    "    best_score = np.max(scores)\n",
    "    min_score = np.min(scores_suggested_sequences)\n",
    "    mean_score = np.mean(scores_suggested_sequences)\n",
    "    print(f'Best GB1 found so far: {best_seq} / {best_score:.3f} (min: {min_score:.3f}, mean: {mean_score:.3f}, max: {best_score:.3f})')\n",
    "    print('')\n",
    "    \n",
    "    df = pd.DataFrame(data=data, columns=('iter', 'sequence', 'exp_value'))\n",
    "    df.to_csv('results_gb1_homolog.csv', index=False)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d9b37c43-cebb-4f56-bb21-393cfb946145",
   "metadata": {},
   "source": [
    "### With ESM-1b and ESM-IF1 (T=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "c19d0a69-2db6-46be-b33c-79c6d3c0706a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load protein language model\n",
    "plm = ProteinEmbedding(pretrained_model_name='esm1b_t33_650M_UR50S', embedding_type='avg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "e80bee52-978c-43ec-ac5d-acef12d02483",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/scicore/home/schwede/eberha0001/Applications/mambaforge/envs/mobius-ubuntu/lib/python3.11/site-packages/esm/pretrained.py:215: UserWarning: Regression weights not found, predicting contacts will not produce correct results.\n",
      "  warnings.warn(\n"
     ]
    }
   ],
   "source": [
    "# Get probabilities from structure model\n",
    "iv = InverseFolding()\n",
    "probabilities = iv.get_probabilities_from_structure('2gi9.pdb', target_chainid='A', temperature=0.5)\n",
    "monomers = iv.vocab\n",
    "\n",
    "# Fix positions involved in the fluorophore activity\n",
    "wt_seq = 'MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGVDGEWTYDDATKTFTVTE'\n",
    "positions = [39, 40, 41, 54]\n",
    "fixed_positions = {}\n",
    "\n",
    "for i, aa in enumerate(wt_seq):\n",
    "    if not i + 1 in positions:\n",
    "        fixed_positions[i + 1] = aa\n",
    "\n",
    "with open('design_protocol_gb1_probabilities.yaml', 'w') as w:\n",
    "    data = generate_biopolymer_design_protocol_from_probabilities(probabilities, monomers, fixed_positions=fixed_positions)\n",
    "    yaml.dump(data, w, sort_keys=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "45f708bf-2ce1-497f-9c15-b4b186debde0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load surrogate model\n",
    "gpmodel = GPLLModel(kernel=RBFKernel(), pretrained_model=plm, noise_prior=NormalPrior(0, 1))\n",
    "ei = ExpectedImprovement(gpmodel, maximize=True)\n",
    "optimizer = SequenceGA(algorithm='GA', period=15, design_protocol_filename='design_protocol_gb1_probabilities.yaml')\n",
    "ps = Planner(ei, optimizer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "abf8ee50-4409-4578-98c5-4d52b0740382",
   "metadata": {},
   "outputs": [],
   "source": [
    "lead_protein = convert_FASTA_to_HELM('MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGVDGEWTYDDATKTFTVTE')[0]\n",
    "\n",
    "seed_library = [lead_protein]\n",
    "positions = {'PEPTIDE1': [39, 40, 41, 54]}\n",
    "\n",
    "for seq in homolog_scanning(lead_protein, positions=positions):\n",
    "    seed_library.append(seq)\n",
    "\n",
    "    if len(seed_library) >= 96:\n",
    "        print('Reach max. number of protein allowed.')\n",
    "        break\n",
    "\n",
    "seed_library = convert_HELM_to_FASTA(seed_library)\n",
    "scores_seed_library = gb1.score(seed_library)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "d344a8e5-4953-49f2-90ee-2d27d35691a1",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Fitting GPLLM model (cuda): 14step [00:16,  1.15s/step, loss=1.02]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==========================================================================================================\n",
      "n_gen  |  n_eval  |        f_avg         |        f_min         |              best sequence              \n",
      "==========================================================================================================\n",
      "     1 |        0 |  7.15435665664993881 |  3.89803670068332009 | MQYKLILNGKTLKGETTT....VWGEWTYDDATKTFTVTE\n",
      "     2 |      500 | -5.3339713128006E-04 | -4.7858195727885E-02 | MQYKLILNGKTLKGETTT....IYGEWTYDDATKTFTVTE\n",
      "     3 |     1000 | -9.5084000093757E-04 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     4 |     1500 | -1.1663151690592E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     5 |     2000 | -1.4781618335290E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     6 |     2500 | -1.6558808208940E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     7 |     3000 | -1.7973118277524E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     8 |     3500 | -1.8959649733395E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "     9 |     4000 | -1.9996499133466E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    10 |     4500 | -2.0785788604474E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    11 |     5000 | -2.1400521467122E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    12 |     5500 | -2.1831559620954E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    13 |     6000 | -2.2084694527735E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    14 |     6500 | -2.2177586433150E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    15 |     7000 | -2.2206158818492E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    16 |     7500 | -2.2216177885403E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    17 |     8000 | -2.2239178574713E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "    18 |     8500 | -2.2256457361617E-03 | -6.3728547093237E-02 | MQYKLILNGKTLKGETTT....VYGEWTYDDATKTFTVTE\n",
      "Best GB1 found so far: MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGIWGEWTYDDATKTFTFTE / 5.772 (min: 0.000, mean: 1.524, max: 5.772)\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Fitting GPLLM model (cuda): 11step [00:19,  1.82s/step, loss=1.26]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==========================================================================================================\n",
      "n_gen  |  n_eval  |        f_avg         |        f_min         |              best sequence              \n",
      "==========================================================================================================\n",
      "     1 |        0 |  1.0411398689636E+01 |  5.77203220933999983 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     2 |      500 | -4.6405833316381E-05 | -3.0533656576523E-03 | MQYKLILNGKTLKGETTT....FWGEWTYDDATKTFTFTE\n",
      "     3 |     1000 | -1.6206212542011E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     4 |     1500 | -3.0681237879490E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     5 |     2000 | -5.0595525598348E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     6 |     2500 | -6.4054309219427E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     7 |     3000 | -6.9276497381340E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     8 |     3500 | -7.2242942753764E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "     9 |     4000 | -7.3764078478724E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "    10 |     4500 | -7.3993600957181E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "    11 |     5000 | -7.4177689982223E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "    12 |     5500 | -7.4268797104342E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "    13 |     6000 | -7.4325900226053E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "    14 |     6500 | -7.4325900226053E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "    15 |     7000 | -7.4337285992285E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "    16 |     7500 | -7.4356533663105E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "    17 |     8000 | -7.4363305093686E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "    18 |     8500 | -7.4363305093686E-04 | -1.2999269784332E-02 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTWTE\n",
      "Best GB1 found so far: MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGIWGEWTYDDATKTFTFTE / 5.772 (min: 0.000, mean: 1.240, max: 5.772)\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Fitting GPLLM model (cuda): 11step [00:41,  3.80s/step, loss=1.19]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==========================================================================================================\n",
      "n_gen  |  n_eval  |        f_avg         |        f_min         |              best sequence              \n",
      "==========================================================================================================\n",
      "     1 |        0 |  1.0373018931908E+01 |  5.77203220933999983 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     2 |      500 | -8.8218575938857E-07 | -1.0299009750796E-04 | MQYKLILNGKTLKGETTT....TYGEWTYDDATKTFTCTE\n",
      "     3 |     1000 | -3.7627246546606E-06 | -2.8842208775472E-04 | MQYKLILNGKTLKGETTT....FYGEWTYDDATKTFTCTE\n",
      "     4 |     1500 | -1.0490789278710E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     5 |     2000 | -1.6338334320422E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     6 |     2500 | -2.1452949250834E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     7 |     3000 | -2.6654403851153E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     8 |     3500 | -3.0614365472894E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     9 |     4000 | -3.3218877982751E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "    10 |     4500 | -3.4788966788826E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "    11 |     5000 | -3.5714734725396E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "    12 |     5500 | -3.6104270555797E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "    13 |     6000 | -3.6542755298889E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "    14 |     6500 | -3.6699671904367E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "    15 |     7000 | -3.6827453517341E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "    16 |     7500 | -3.6945627576497E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "    17 |     8000 | -3.6978671475652E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "    18 |     8500 | -3.7017590566887E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "    19 |     9000 | -3.7021428516193E-05 | -1.0459754492522E-03 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "Best GB1 found so far: MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGIWGEWTYDDATKTFTFTE / 5.772 (min: 0.000, mean: 1.406, max: 5.772)\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Fitting GPLLM model (cuda): 14step [01:40,  7.18s/step, loss=1.24]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==========================================================================================================\n",
      "n_gen  |  n_eval  |        f_avg         |        f_min         |              best sequence              \n",
      "==========================================================================================================\n",
      "     1 |        0 |  1.0311340966618E+01 |  5.77203220933999983 | MQYKLILNGKTLKGETTT....IWGEWTYDDATKTFTFTE\n",
      "     2 |      500 | -9.8627169848210E-07 | -8.6810815067428E-05 | MQYKLILNGKTLKGETTT....LRCEWTYDDATKTFTGTE\n",
      "     3 |     1000 | -9.3558381645971E-06 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "     4 |     1500 | -1.3868806297552E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "     5 |     2000 | -1.7545127476574E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "     6 |     2500 | -2.0235069350704E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "     7 |     3000 | -2.2307056071480E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "     8 |     3500 | -2.3852981387018E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "     9 |     4000 | -2.5314587857969E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "    10 |     4500 | -2.7648268139219E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "    11 |     5000 | -2.8682943680142E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "    12 |     5500 | -2.9380885536846E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "    13 |     6000 | -3.1022634816787E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "    14 |     6500 | -3.1553410552282E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "    15 |     7000 | -3.2004437122071E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "    16 |     7500 | -3.2158596042550E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "    17 |     8000 | -3.2350063206874E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "    18 |     8500 | -3.2430415885007E-05 | -1.2629101543135E-03 | MQYKLILNGKTLKGETTT....THCEWTYDDATKTFTATE\n",
      "Best GB1 found so far: MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGAHCEWTYDDATKTFTATE / 9.913 (min: 0.002, mean: 1.600, max: 9.913)\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Fitting GPLLM model (cuda): 13step [01:38,  7.55s/step, loss=1.38]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==========================================================================================================\n",
      "n_gen  |  n_eval  |        f_avg         |        f_min         |              best sequence              \n",
      "==========================================================================================================\n",
      "     1 |        0 |  1.8517001510241E+01 |  9.91308832738880064 | MQYKLILNGKTLKGETTT....AHCEWTYDDATKTFTATE\n",
      "     2 |      500 | -7.0897978823014E-13 | -2.4700570655305E-10 | MQYKLILNGKTLKGETTT....WWCEWTYDDATKTFTGTE\n",
      "     3 |     1000 | -1.9250064671620E-12 | -4.0064413493086E-10 | MQYKLILNGKTLKGETTT....ANCEWTYDDATKTFTATE\n",
      "     4 |     1500 | -1.7841975641506E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "     5 |     2000 | -2.2801034233812E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "     6 |     2500 | -2.9552371732440E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "     7 |     3000 | -3.0375144728455E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "     8 |     3500 | -3.1855525354993E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "     9 |     4000 | -3.2780507190189E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "    10 |     4500 | -3.3214707387659E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "    11 |     5000 | -3.3563189557228E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "    12 |     5500 | -3.4168985793972E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "    13 |     6000 | -3.4841977218351E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "    14 |     6500 | -3.5209440279333E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "    15 |     7000 | -3.5508042645301E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "    16 |     7500 | -3.6233487539991E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "    17 |     8000 | -3.6423074666547E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "    18 |     8500 | -3.6533160937865E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "    19 |     9000 | -3.6736012345213E-11 | -4.2092131042628E-09 | MQYKLILNGKTLKGETTT....ASCEWTYDDATKTFTATE\n",
      "Best GB1 found so far: MQYKLILNGKTLKGETTTEAVDAATAEKVFKQYANDNGAHCEWTYDDATKTFTATE / 9.913 (min: 0.000, mean: 2.115, max: 9.913)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "sequences = seed_library.copy()\n",
    "scores = scores_seed_library.copy()\n",
    "        \n",
    "# Store data for later analysis\n",
    "data = [(0, p, s) for p, s in zip(sequences, scores)]\n",
    "        \n",
    "for i in range(5):\n",
    "    suggested_sequences, _ = ps.recommand(sequences, scores.reshape(-1, 1), batch_size=96)\n",
    "    scores_suggested_sequences = gb1.score(suggested_sequences)\n",
    "        \n",
    "    sequences = np.concatenate([sequences, suggested_sequences])\n",
    "    scores = np.concatenate((scores, scores_suggested_sequences), axis=0)\n",
    "    data.extend([(i + 1, p, s) for p, s in zip(suggested_sequences, scores_suggested_sequences)])\n",
    "        \n",
    "    best_seq = sequences[np.argmax(scores)]\n",
    "    best_score = np.max(scores)\n",
    "    min_score = np.min(scores_suggested_sequences)\n",
    "    mean_score = np.mean(scores_suggested_sequences)\n",
    "    print(f'Best GB1 found so far: {best_seq} / {best_score:.3f} (min: {min_score:.3f}, mean: {mean_score:.3f}, max: {best_score:.3f})')\n",
    "    print('')\n",
    "    \n",
    "    df = pd.DataFrame(data=data, columns=('iter', 'sequence', 'exp_value'))\n",
    "    df.to_csv('results_gb1_homolog_esm-if1.csv', index=False)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0134c6f1-6dab-4cee-ae0c-ecdaa87d275a",
   "metadata": {},
   "source": [
    "## Comparison"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "eb717429-d1b3-450c-bf81-6fe971800a99",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "methods = ['gb1_homolog', 'gb1_homolog_esm-if1']\n",
    "method_renaming = {'gb1_homolog': 'Homolog + ESM-1b', 'gb1_homolog_esm-if1': 'Homolog + ESM-1b + ESM-IF1 (T=0.5)'}\n",
    "\n",
    "dfs = []\n",
    "\n",
    "for method in methods:\n",
    "    df = pd.read_csv(f'results_{method}.csv')\n",
    "    df['method'] = method_renaming[method]\n",
    "    maxs = df.loc[df.groupby(by=['iter'])['exp_value'].idxmax()]\n",
    "    dfs.append(maxs)\n",
    "\n",
    "dfs = pd.concat(dfs).reset_index()\n",
    "\n",
    "dfs['iter'] = dfs['iter'].replace({0: 'Init.', 1: '1', 2: '2', 3: '3', 4: '4', 5: '5'})\n",
    "\n",
    "fig, axarr = plt.subplots(figsize=(10, 5))\n",
    "sns.lineplot(x='iter', y='exp_value', hue='method', data=dfs, ax=axarr, linewidth=3.5)\n",
    "\n",
    "axarr.legend(loc='upper right', fontsize=15, frameon=False)\n",
    "axarr.set_ylim([-0.1, 10.1])\n",
    "axarr.set_xlabel(\"Generations\", fontsize=15)\n",
    "axarr.set_ylabel(\"GB1 Binding Fitness\", fontsize=15)\n",
    "axarr.xaxis.set_tick_params(labelsize=15)\n",
    "axarr.yaxis.set_tick_params(labelsize=15)\n",
    "\n",
    "axarr.get_legend().remove()\n",
    "axarr.legend(bbox_to_anchor=(0., 1., 1., 0.2), loc=3, ncol=2, mode=\"expand\", borderaxespad=0, fontsize=15, frameon=False)\n",
    "\n",
    "sns.despine()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "07c64f2c-793a-4c19-ac3e-f3912ec45076",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "methods = ['gb1_homolog', 'gb1_homolog_esm-if1']\n",
    "method_renaming = {'gb1_homolog': 'Homolog + ESM-1b', 'gb1_homolog_esm-if1': 'Homolog + ESM-1b + ESM-IF1 (T=0.5)'}\n",
    "\n",
    "dfs = []\n",
    "\n",
    "for method in methods:\n",
    "    df = pd.read_csv(f'results_{method}.csv')\n",
    "    df['method'] = method_renaming[method]\n",
    "    dfs.append(df)\n",
    "\n",
    "dfs = pd.concat(dfs).reset_index()\n",
    "\n",
    "dfs['iter'] = dfs['iter'].replace({0: 'Init.', 1: '1', 2: '2', 3: '3', 4: '4', 5: '5'})\n",
    "\n",
    "fig, axarr = plt.subplots(figsize=(10, 5))\n",
    "sns.lineplot(x='iter', y='exp_value', hue='method', data=dfs, ax=axarr, linewidth=3.5)\n",
    "\n",
    "axarr.legend(loc='upper right', fontsize=15, frameon=False)\n",
    "axarr.set_ylim([-0.1, 10.1])\n",
    "axarr.set_xlabel(\"Generations\", fontsize=15)\n",
    "axarr.set_ylabel(\"GB1 Binding Fitness\", fontsize=15)\n",
    "axarr.xaxis.set_tick_params(labelsize=15)\n",
    "axarr.yaxis.set_tick_params(labelsize=15)\n",
    "\n",
    "axarr.get_legend().remove()\n",
    "axarr.legend(bbox_to_anchor=(0., 1., 1., 0.2), loc=3, ncol=2, mode=\"expand\", borderaxespad=0, fontsize=15, frameon=False)\n",
    "\n",
    "sns.despine()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4e751c13-32ef-4e12-9575-6001975d849f",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:mambaforge-mobius-ubuntu]",
   "language": "python",
   "name": "conda-env-mambaforge-mobius-ubuntu-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}