Bio2ByteTools_v3_singleseq_pypi.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "PIP - B2B Tools",
      "provenance": [],
      "collapsed_sections": [],
      "private_outputs": true,
      "machine_shape": "hm",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Bio2Byte/public_notebooks/blob/main/Bio2ByteTools_v3_singleseq_pypi.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Bio2Byte - Single Sequence Predictions\n",
        "For most of below tools, we now have the b2bTools one-stop file upload location where you can easily try them: [online b2BTools](https://bio2byte.be/b2btools/)"
      ],
      "metadata": {
        "id": "EsZPZMEGYV2d"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "YaxPffRnrSVW",
        "cellView": "form"
      },
      "outputs": [],
      "source": [
        "#@title Install dependencies\n",
        "%%capture\n",
        "\n",
        "!pip install b2bTools==3.0.5"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Import dependencies\n",
        "import os\n",
        "import json\n",
        "from google.colab import files\n",
        "from b2bTools import SingleSeq\n",
        "from matplotlib import pyplot as plt\n",
        "from matplotlib.lines import Line2D\n",
        "\n",
        "if not os.path.exists(\"/content/results\"):\n",
        "    os.mkdir('/content/results')"
      ],
      "metadata": {
        "id": "cKwsuaFgr3CT",
        "cellView": "form"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Input protein sequences\n",
        "%%capture\n",
        "\n",
        "query_sequence_1 = 'YRRRRYYYYYRRRYYYYYYYMESNQSNNGGSGRYYYYYYYNAALNRGGRYVPPHLRGGDGGAAAAASAGGDDRRGGAGGGGYRRGGGNSGGGGGGGYDRGYNDNRDDRDNRGGSGGYGRDRNYEDRGYNGGGGGGGNRGYNNNRGGGGGGYNRQDRGDGGSSNFSRGGYNNRDEGSDNRGSGRSYNNDRRDNGGDGQNTRWNNLDAPPSRGTSKWENRGARDERIEQELFSGQLSGINFDKYEEIPVEATGDDVPQPISLFSDLSLHE' #@param {type:\"string\"}\n",
        "query_sequence_2 = '' #@param {type:\"string\"}\n",
        "query_sequence_3 = '' #@param {type:\"string\"}\n",
        "query_sequence_4 = '' #@param {type:\"string\"}\n",
        "query_sequence_5 = '' #@param {type:\"string\"}\n",
        "query_sequence_6 = '' #@param {type:\"string\"}\n",
        "query_sequence_7 = '' #@param {type:\"string\"}\n",
        "query_sequence_8 = '' #@param {type:\"string\"}\n",
        "query_sequence_9 = '' #@param {type:\"string\"}\n",
        "query_sequence_10 = '' #@param {type:\"string\"}\n",
        "\n",
        "fasta = f'>SEQ_1\\n{query_sequence_1}\\n'\n",
        "\n",
        "if query_sequence_2 != '':\n",
        "    fasta += f'>SEQ_2\\n{query_sequence_2}\\n'\n",
        "if query_sequence_3 != '':\n",
        "    fasta += f'>SEQ_3\\n{query_sequence_3}\\n'\n",
        "if query_sequence_4 != '':\n",
        "    fasta += f'>SEQ_4\\n{query_sequence_4}\\n'\n",
        "if query_sequence_5 != '':\n",
        "    fasta += f'>SEQ_5\\n{query_sequence_5}\\n'\n",
        "if query_sequence_6 != '':\n",
        "    fasta += f'>SEQ_6\\n{query_sequence_6}\\n'\n",
        "if query_sequence_7 != '':\n",
        "    fasta += f'>SEQ_7\\n{query_sequence_7}\\n'\n",
        "if query_sequence_8 != '':\n",
        "    fasta += f'>SEQ_8\\n{query_sequence_8}\\n'\n",
        "if query_sequence_9 != '':\n",
        "    fasta += f'>SEQ_9\\n{query_sequence_9}\\n'\n",
        "if query_sequence_10 != '':\n",
        "    fasta += f'>SEQ_10\\n{query_sequence_10}\\n'\n",
        "\n",
        "with open(\"input_example1.fasta\", \"w\") as f:\n",
        "  f.write(fasta)\n",
        "\n",
        "single_seq = SingleSeq(\"/content/input_example1.fasta\")\n",
        "single_seq.predict(tools=['dynamine', 'efoldmine', 'disomine', 'agmata', 'psp'])\n",
        "\n",
        "all_predictions = single_seq.get_all_predictions()\n",
        "keys = list(all_predictions.keys())\n",
        "\n",
        "max_seq_len = max([len(pred['seq']) for pred in all_predictions.values()])"
      ],
      "metadata": {
        "id": "a4AYPR-ByvL4",
        "cellView": "form"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Plot Single Sequence predictions\n",
        "#@markdown The predictions reflect 'emerging' properties, so what the sequence \n",
        "#@markdown is capable of, not necessarily what it will do in a particular \n",
        "#@markdown context, for example when it adopts a specific fold.\n",
        "\n",
        "#@markdown #### DynaMine backbone dynamics\n",
        "#@markdown > Values above 0.8 indicate rigid conformations, \n",
        "#@markdown > values above 1.0 membrane spanning regions, \n",
        "#@markdown > values below 0.69 flexible regions. \n",
        "#@markdown > Values between 0.69-0.80 are 'context' dependent and capable of\n",
        "#@markdown > being either rigid or flexible.\n",
        "\n",
        "#@markdown #### DynaMine sidechain dynamics\n",
        "#@markdown > Higher values mean more likely rigid. \n",
        "#@markdown > These values are highly dependent on the amino acid type \n",
        "#@markdown > (i.e. a Trp will be rigid, an Asp flexible).\n",
        "\n",
        "#@markdown #### DynaMine conformational propensities (sheet, helix, coil, ppII (polyproline II))\n",
        "#@markdown > Higher values indicate higher propensities.\n",
        "\n",
        "#@markdown #### EFoldMine earlyFolding propensity\n",
        "#@markdown > Values above 0.169 indicate residues that are likely to start \n",
        "#@markdown > the protein folding process, based on only local interactions \n",
        "#@markdown > with other amino acids.\n",
        "\n",
        "#@markdown #### Disomine disorder\n",
        "#@markdown >  Values above 0.5 indicate that this is likely a disordered residue.\n",
        "\n",
        "\n",
        "fig, axs = plt.subplots(2, 4)\n",
        "\n",
        "ax1 = axs[0, 0]\n",
        "ax2 = axs[0, 1]\n",
        "ax3 = axs[0, 2]\n",
        "ax4 = axs[0, 3]\n",
        "ax5 = axs[1, 0]\n",
        "ax6 = axs[1, 1]\n",
        "ax7 = axs[1, 2]\n",
        "ax8 = axs[1, 3]\n",
        "\n",
        "fig.set_figwidth(30)\n",
        "fig.set_figheight(10)\n",
        "\n",
        "fig.suptitle(f'Single Sequence Predictions')\n",
        "\n",
        "for seq_key in keys:\n",
        "    predictions = all_predictions[seq_key]\n",
        "    x_position = range(len(predictions['seq']))\n",
        "\n",
        "    backbone_pred = predictions['backbone']\n",
        "    coil_pred = predictions['coil']\n",
        "    sheet_pred = predictions['sheet']\n",
        "    ppII_pred = predictions['ppII']\n",
        "    helix_pred = predictions['helix']\n",
        "    sidechain_pred = predictions['sidechain']\n",
        "    disomine_pred = predictions['disoMine']\n",
        "    earlyFolding_pred = predictions['earlyFolding'] \n",
        "    agmata_pred = predictions['agmata']\n",
        "\n",
        "    ax1.plot(x_position, backbone_pred, label=seq_key)\n",
        "    ax2.plot(x_position, sidechain_pred, label=seq_key)\n",
        "    ax3.plot(x_position, coil_pred, label=seq_key)\n",
        "    ax4.plot(x_position, sheet_pred, label=seq_key)\n",
        "    ax5.plot(x_position, ppII_pred, label=seq_key)\n",
        "    ax6.plot(x_position, helix_pred, label=seq_key)\n",
        "    ax7.plot(x_position, disomine_pred, label=seq_key)\n",
        "    ax8.plot(x_position, earlyFolding_pred, label=seq_key)\n",
        "\n",
        "ax1.set_title('DynaMine backbone dynamics')\n",
        "ax1.set_ylim([-0.2, 1.2])\n",
        "ax1.set_xlabel('residue index')\n",
        "ax1.set_ylabel('prediction values')\n",
        "ax1.axhspan(1, 1.2, alpha=0.3, color='red')\n",
        "ax1.axhspan(0.8, 1, alpha=0.5, color='pink')\n",
        "ax1.axhspan(0.69, 0.8, alpha=0.5, color='orange')\n",
        "ax1.axhspan(-0.2, 0.69, alpha=0.5, color='yellow')\n",
        "ax1.grid(axis='y')\n",
        "ax1.set_xlim([0, max_seq_len - 1])\n",
        "\n",
        "\n",
        "ax2.set_title('DynaMine sidechain dynamics')\n",
        "ax2.set_ylim([-0.2, 1.2])\n",
        "ax2.set_xlabel('residue index')\n",
        "ax2.set_ylabel('prediction values')\n",
        "ax2.grid(axis='y')\n",
        "ax2.set_xlim([0, max_seq_len - 1])\n",
        "\n",
        "\n",
        "ax3.set_title('DynaMine conformational propensities: Coil')\n",
        "ax3.set_ylim([-0.2, 1.2])\n",
        "ax3.set_xlabel('residue index')\n",
        "ax3.set_ylabel('prediction values')\n",
        "ax3.grid(axis='y')\n",
        "ax3.set_xlim([0, max_seq_len - 1])\n",
        "\n",
        "\n",
        "ax4.set_title('DynaMine conformational propensities: Sheet')\n",
        "ax4.set_ylim([-0.2, 1.2])\n",
        "ax4.set_xlabel('residue index')\n",
        "ax4.set_ylabel('prediction values')\n",
        "ax4.grid(axis='y')\n",
        "ax4.set_xlim([0, max_seq_len - 1])\n",
        "\n",
        "ax5.set_title('DynaMine conformational propensities: ppII (polyproline II)')\n",
        "ax5.set_ylim([-0.2, 1.2])\n",
        "ax5.set_xlabel('residue index')\n",
        "ax5.set_ylabel('prediction values')\n",
        "ax5.grid(axis='y')\n",
        "ax5.set_xlim([0, max_seq_len - 1])\n",
        "\n",
        "\n",
        "ax6.set_title('DynaMine conformational propensities: Helix')\n",
        "ax6.set_ylim([-0.2, 1.2])\n",
        "ax6.set_xlabel('residue index')\n",
        "ax6.set_ylabel('prediction values')\n",
        "ax6.grid(axis='y')\n",
        "ax6.set_xlim([0, max_seq_len - 1])\n",
        "\n",
        "\n",
        "ax7.set_title('Early folding (EFoldMine)')\n",
        "ax7.set_ylim([-0.2, 1.2])\n",
        "ax7.set_xlabel('residue index')\n",
        "ax7.set_ylabel('prediction values')\n",
        "ax7.axhspan(-0.2, 0.169, alpha=0.5, color='yellow')\n",
        "ax7.axhspan(0.169, 1.2, alpha=0.5, color='orange')\n",
        "ax7.grid(axis='y')\n",
        "ax7.set_xlim([0, max_seq_len - 1])\n",
        "\n",
        "\n",
        "ax8.set_title('Disorder (disoMine)')\n",
        "ax8.set_ylim([-0.2, 1.2])\n",
        "ax8.set_xlabel('residue index')\n",
        "ax8.set_ylabel('prediction values')\n",
        "ax8.axhspan(0.5, 1.2, alpha=0.5, color='orange')\n",
        "ax8.axhspan(-0.2, 0.5, alpha=0.5, color='yellow')\n",
        "ax8.grid(axis='y')\n",
        "ax8.set_xlim([0, max_seq_len - 1])\n",
        "\n",
        "plt.legend(loc='upper center', bbox_to_anchor=(0.5, -0.3), fancybox=True, shadow=True, ncol=len(keys))\n",
        "plt.subplots_adjust(hspace=0.4)\n",
        "plt.savefig('/content/results/single_sequence_predictons.png')\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "KOH6FzXFrzw4",
        "cellView": "form"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Plot AgMata aggregation propensity\n",
        "#@markdown The predictions reflect 'emerging' properties, so what the sequence \n",
        "#@markdown is capable of, not necessarily what it will do in a particular \n",
        "#@markdown context, for example when it adopts a specific fold.\n",
        "\n",
        "#@markdown > These values are divided by a factor of 20 from the original. \n",
        "#@markdown > Peaks indicate residues likely to be involved in beta-sheet \n",
        "#@markdown > aggregation.\n",
        "\n",
        "fig, ax = plt.subplots(1, 1)\n",
        "fig.set_figwidth(30)\n",
        "fig.set_figheight(5)\n",
        "fig.suptitle('Agmata aggregation propensity')\n",
        "\n",
        "for seq_key in keys:\n",
        "    predictions = all_predictions[seq_key]\n",
        "    agmata_pred = predictions['agmata']\n",
        "    ax.plot(range(len(agmata_pred)), agmata_pred, label=seq_key)\n",
        "\n",
        "ax.set_xlim([0, max_seq_len - 1])\n",
        "ax.set_xlabel('residue index')\n",
        "ax.set_ylabel('prediction values')\n",
        "ax.grid(axis='y')\n",
        "\n",
        "plt.legend(loc='upper center', bbox_to_anchor=(0.5, -0.2), fancybox=True, shadow=True, ncol=len(keys))\n",
        "plt.savefig('/content/results/agmata_predictons.png')\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "m_K9OQKnSXg3",
        "cellView": "form"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Plot PSPer\n",
        "#@markdown PSP (Phase Separating Protein) enables you to predict whether a \n",
        "#@markdown protein is likely to phase-separate with a particular mechanism \n",
        "#@markdown involving RNA interacts (FUS-like proteins). \n",
        "#@markdown It will highlight the regions in your protein that are involved \n",
        "#@markdown mechanistically, and provide an overall score.\n",
        "\n",
        "for seq_key in keys:\n",
        "    predictions = all_predictions[seq_key]\n",
        "    sequence = predictions['seq']\n",
        "    x = range(len(sequence))\n",
        "\n",
        "    tyr_x = []\n",
        "    tyr_y = []\n",
        "    arg_x = []\n",
        "    arg_y = []\n",
        "    \n",
        "    for ind, residue in enumerate(sequence):\n",
        "        if residue == 'Y':\n",
        "            tyr_x.append(ind)\n",
        "            tyr_y.append(0.5)\n",
        "        if residue == 'R':\n",
        "            arg_x.append(ind)\n",
        "            arg_y.append(0.5)\n",
        "\n",
        "    viterbi_path = predictions['viterbi']\n",
        "    color_map = lambda step: '#ffb499' if step == ' RRM' else '#9999ff' if step == ' PLD' else '#bfbfbf' if step == ' SPACER' else '#99ff99' if step == ' OTHER' else '#fff'\n",
        "    viterbi_colors = list(map(color_map, viterbi_path))\n",
        "\n",
        "    fig, (ax1, ax2, ax3, ax4, ax5, ax6) = plt.subplots(6, 1, gridspec_kw={'height_ratios': [1, 3, 3, 3, 3, 3]})\n",
        "    fig.set_figwidth(30)\n",
        "    fig.set_figheight(15)\n",
        "    fig.suptitle(f'PSPer {seq_key} (PSPer score: {predictions[\"protein_score\"]:2.3}, length {len(sequence)} AA)', size=14, y=1.05)\n",
        "    \n",
        "    # AX1\n",
        "    if predictions[\"protein_score\"] >= 0.5611191832364677:\n",
        "        ax1.set_title(\"Likely a phase-separation protein\", loc='left')\n",
        "    else:\n",
        "        ax1.set_title(\"Likely not a phase-separation protein\", loc='left')\n",
        "\n",
        "    ax1.set_xlim([0, len(sequence) - 1])\n",
        "    ax1.set_ylim([0, 1])\n",
        "    ax1.scatter(tyr_x, tyr_y, marker=\"o\", color=\"blue\")\n",
        "    ax1.scatter(arg_x, arg_y, marker=\"o\", color=\"red\")\n",
        "    ax1.axes.yaxis.set_visible(False)\n",
        "    ax1.set_xlabel('residue index')\n",
        "    \n",
        "    # Complexity\n",
        "    ax2.plot(x, predictions['complexity'], color='#808080')\n",
        "    ax2.set_title('Complexity', loc='left')\n",
        "    ax2.set_ylabel('Seq. complexity')\n",
        "    ax2.set_xlabel('residue index')\n",
        "    ax2.set_xlim([0, len(sequence) - 1])\n",
        "    ax2.set_ylim([0, max(predictions['complexity']) + 1])\n",
        "    ax2.grid()\n",
        "\n",
        "    # Tyr Enrichment\n",
        "    ax3.plot(x, predictions['tyr'], color='#808080')\n",
        "    ax3.set_title('Tyr', loc='left')\n",
        "    ax3.set_ylabel('Tyr Enrichment')\n",
        "    ax3.set_xlabel('residue index')\n",
        "    ax3.set_xlim([0, len(sequence) - 1])\n",
        "    ax3.set_ylim([0, max(predictions['tyr']) + 1])\n",
        "    ax3.grid()\n",
        "\n",
        "    # Arg Enrichment\n",
        "    ax4.plot(x, predictions['arg'], color='#808080')\n",
        "    ax4.set_title('Arg', loc='left')\n",
        "    ax4.set_ylabel('Arg Enrichment')\n",
        "    ax4.set_xlabel('residue index')\n",
        "    ax4.set_xlim([0, len(sequence) - 1])\n",
        "    ax4.set_ylim([0, max(predictions['arg']) + 1])\n",
        "    ax4.grid()\n",
        "\n",
        "    # RRM \n",
        "    ax5.plot(x, predictions['RRM'], color='#808080')\n",
        "    ax5.set_title('RRM', loc='left')\n",
        "    ax5.set_ylabel('RRM')\n",
        "    ax5.set_xlabel('residue index')\n",
        "    ax5.set_xlim([0, len(sequence) - 1])\n",
        "    ax5.grid()\n",
        "\n",
        "    # Disorder\n",
        "    ax6.plot(x, predictions['disorder'], color='#808080')\n",
        "    ax6.set_title('Disorder', loc='left')\n",
        "    ax6.set_ylabel('Disorder')\n",
        "    ax6.set_xlabel('residue index')\n",
        "    ax6.set_xlim([0, len(sequence) - 1])\n",
        "    ax6.set_ylim([0, max(predictions['disorder']) + 1])\n",
        "    ax6.grid()\n",
        "\n",
        "    for i in x:\n",
        "        ax1.axvspan(i, i+1, facecolor=viterbi_colors[i], alpha=0.5)\n",
        "        ax2.axvspan(i, i+1, facecolor=viterbi_colors[i], alpha=0.5)\n",
        "        ax3.axvspan(i, i+1, facecolor=viterbi_colors[i], alpha=0.5)\n",
        "        ax4.axvspan(i, i+1, facecolor=viterbi_colors[i], alpha=0.5)\n",
        "        ax5.axvspan(i, i+1, facecolor=viterbi_colors[i], alpha=0.5)\n",
        "        ax6.axvspan(i, i+1, facecolor=viterbi_colors[i], alpha=0.5)\n",
        "\n",
        "    axbox = ax1.get_position()\n",
        "    \n",
        "    custom_lines = [\n",
        "        Line2D([0], [0], color=\"white\", marker='o', markerfacecolor=\"blue\"),\n",
        "        Line2D([0], [0], color=\"white\", marker='o', markerfacecolor=\"red\"),\n",
        "        Line2D([0], [0], color='#99ff99', lw=3),\n",
        "        Line2D([0], [0], color='#bfbfbf', lw=3),\n",
        "        Line2D([0], [0], color='#ffb499', lw=3),\n",
        "        Line2D([0], [0], color='#9999ff', lw=3)\n",
        "    ]\n",
        "    \n",
        "    fig.legend(\n",
        "        custom_lines, \n",
        "        ['Tyr', 'Arg', 'Other', 'Spacer', 'putative RRM', 'putative PLD'], \n",
        "        loc='center',\n",
        "        ncol=6,\n",
        "        bbox_to_anchor=[axbox.x0 + 0.5*axbox.width, axbox.y1 + 0.12], bbox_transform=fig.transFigure\n",
        "    )\n",
        "    fig.tight_layout()\n",
        "    print(f\"PSPer plot for {seq_key}\")\n",
        "    plt.savefig(f'/content/results/psp_{seq_key}.png')\n",
        "    plt.show()\n",
        "\n"
      ],
      "metadata": {
        "id": "Zk-eH4fI3dSJ",
        "cellView": "form"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Plot Single Sequence predictions (3D projection)\n",
        "#@markdown The predictions reflect 'emerging' properties, so what the sequence \n",
        "#@markdown is capable of, not necessarily what it will do in a particular \n",
        "#@markdown context, for example when it adopts a specific fold.\n",
        "\n",
        "#@markdown #### DynaMine backbone dynamics\n",
        "#@markdown > Values above 0.8 indicate rigid conformations, \n",
        "#@markdown > values above 1.0 membrane spanning regions, \n",
        "#@markdown > values below 0.69 flexible regions. \n",
        "#@markdown > Values between 0.69-0.80 are 'context' dependent and capable of\n",
        "#@markdown > being either rigid or flexible.\n",
        "\n",
        "#@markdown #### DynaMine sidechain dynamics\n",
        "#@markdown > Higher values mean more likely rigid. \n",
        "#@markdown > These values are highly dependent on the amino acid type \n",
        "#@markdown > (i.e. a Trp will be rigid, an Asp flexible).\n",
        "\n",
        "#@markdown #### DynaMine conformational propensities (sheet, helix, coil, ppII (polyproline II))\n",
        "#@markdown > Higher values indicate higher propensities.\n",
        "\n",
        "#@markdown #### EFoldMine earlyFolding propensity\n",
        "#@markdown > Values above 0.169 indicate residues that are likely to start \n",
        "#@markdown > the protein folding process, based on only local interactions \n",
        "#@markdown > with other amino acids.\n",
        "\n",
        "#@markdown #### Disomine disorder\n",
        "#@markdown >  Values above 0.5 indicate that this is likely a disordered residue.\n",
        "\n",
        "fig = plt.figure(figsize=plt.figaspect(0.5))\n",
        "# axs = fig.add_subplot(2, 4, 1, projection='3d')\n",
        "\n",
        "ax1 = fig.add_subplot(4, 2, 1, projection='3d')\n",
        "ax2 = fig.add_subplot(4, 2, 2, projection='3d')\n",
        "ax3 = fig.add_subplot(4, 2, 3, projection='3d')\n",
        "ax4 = fig.add_subplot(4, 2, 4, projection='3d')\n",
        "ax5 = fig.add_subplot(4, 2, 5, projection='3d')\n",
        "ax6 = fig.add_subplot(4, 2, 6, projection='3d')\n",
        "ax7 = fig.add_subplot(4, 2, 7, projection='3d')\n",
        "ax8 = fig.add_subplot(4, 2, 8, projection='3d')\n",
        "\n",
        "fig.set_figwidth(30)\n",
        "fig.set_figheight(40)\n",
        "\n",
        "fig.suptitle(f'Single Sequence Predictions (3D Projection)')\n",
        "\n",
        "for index, seq_key in enumerate(keys):\n",
        "    predictions = all_predictions[seq_key]\n",
        "    x_position = range(len(predictions['seq']))\n",
        "    z_position = [index] * len(predictions['seq'])\n",
        "\n",
        "    backbone_pred = predictions['backbone']\n",
        "    coil_pred = predictions['coil']\n",
        "    sheet_pred = predictions['sheet']\n",
        "    ppII_pred = predictions['ppII']\n",
        "    helix_pred = predictions['helix']\n",
        "    sidechain_pred = predictions['sidechain']\n",
        "    disomine_pred = predictions['disoMine']\n",
        "    earlyFolding_pred = predictions['earlyFolding'] \n",
        "    agmata_pred = predictions['agmata']\n",
        "\n",
        "    ax1.plot(x_position, z_position, backbone_pred, label=seq_key)\n",
        "    ax2.plot(x_position, z_position, sidechain_pred, label=seq_key)\n",
        "    ax3.plot(x_position, z_position, coil_pred, label=seq_key)\n",
        "    ax4.plot(x_position, z_position, sheet_pred, label=seq_key)\n",
        "    ax5.plot(x_position, z_position, ppII_pred, label=seq_key)\n",
        "    ax6.plot(x_position, z_position, helix_pred, label=seq_key)\n",
        "    ax7.plot(x_position, z_position, disomine_pred, label=seq_key)\n",
        "    ax8.plot(x_position, z_position, earlyFolding_pred, label=seq_key)\n",
        "\n",
        "ax1.set_title('DynaMine backbone dynamics')\n",
        "ax1.set_zlim([-0.1, 1.1])\n",
        "ax1.set_xlabel('residue index')\n",
        "ax1.set_zlabel('prediction values')\n",
        "ax1.set_yticks(range(len(keys)))\n",
        "ax1.set_yticklabels(keys)\n",
        "ax1.view_init(60, -45)\n",
        "\n",
        "\n",
        "ax2.set_title('DynaMine sidechain dynamics')\n",
        "ax2.set_zlim([-0.1, 1.1])\n",
        "ax2.set_xlabel('residue index')\n",
        "ax2.set_zlabel('prediction values')\n",
        "ax2.set_yticks(range(len(keys)))\n",
        "ax2.set_yticklabels(keys)\n",
        "ax2.view_init(60, -45)\n",
        "\n",
        "\n",
        "ax3.set_title('DynaMine conformational propensities: Coil')\n",
        "ax3.set_zlim([-0.1, 1.1])\n",
        "ax3.set_xlabel('residue index')\n",
        "ax3.set_zlabel('prediction values')\n",
        "ax3.set_yticks(range(len(keys)))\n",
        "ax3.set_yticklabels(keys)\n",
        "ax3.view_init(60, -45)\n",
        "\n",
        "\n",
        "ax4.set_title('DynaMine conformational propensities: Sheet')\n",
        "ax4.set_zlim([-0.1, 1.1])\n",
        "ax4.set_xlabel('residue index')\n",
        "ax4.set_zlabel('prediction values')\n",
        "ax4.set_yticks(range(len(keys)))\n",
        "ax4.set_yticklabels(keys)\n",
        "ax4.view_init(60, -45)\n",
        "\n",
        "\n",
        "ax5.set_title('DynaMine conformational propensities: ppII (polyproline II)')\n",
        "ax5.set_zlim([-0.1, 1.1])\n",
        "ax5.set_xlabel('residue index')\n",
        "ax5.set_zlabel('prediction values')\n",
        "ax5.set_yticks(range(len(keys)))\n",
        "ax5.set_yticklabels(keys)\n",
        "ax5.view_init(60, -45)\n",
        "\n",
        "\n",
        "ax6.set_title('DynaMine conformational propensities: Helix')\n",
        "ax6.set_zlim([-0.1, 1.1])\n",
        "ax6.set_xlabel('residue index')\n",
        "ax6.set_ylabel('sequence')\n",
        "ax6.set_zlabel('prediction values')\n",
        "ax6.set_yticks(range(len(keys)))\n",
        "ax6.set_yticklabels(keys)\n",
        "ax6.view_init(60, -45)\n",
        "\n",
        "\n",
        "ax7.set_title('Disorder (disoMine)')\n",
        "ax7.set_zlim([-0.1, 1.1])\n",
        "ax7.set_xlabel('residue index')\n",
        "ax7.set_ylabel('sequence')\n",
        "ax7.set_zlabel('prediction values')\n",
        "ax7.set_yticks(range(len(keys)))\n",
        "ax7.set_yticklabels(keys)\n",
        "ax7.view_init(60, -45)\n",
        "\n",
        "\n",
        "ax8.set_title('Early folding (EFoldMine)')\n",
        "ax8.set_zlim([-0.1, 1.1])\n",
        "ax8.set_xlabel('residue index')\n",
        "ax8.set_ylabel('sequence')\n",
        "ax8.set_zlabel('prediction values')\n",
        "ax8.set_yticks(range(len(keys)))\n",
        "ax8.set_yticklabels(keys)\n",
        "ax8.view_init(60, -45)\n",
        "\n",
        "\n",
        "plt.subplots_adjust(hspace=0.2)\n",
        "plt.savefig('/content/results/single_sequence_predictons_3d.png')\n",
        "plt.show()"
      ],
      "metadata": {
        "cellView": "form",
        "id": "JKVklmzvY9Y3"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Plot AgMata aggregation propensity (3D projection)\n",
        "#@markdown The predictions reflect 'emerging' properties, so what the sequence \n",
        "#@markdown is capable of, not necessarily what it will do in a particular \n",
        "#@markdown context, for example when it adopts a specific fold.\n",
        "\n",
        "#@markdown > These values are divided by a factor of 20 from the original. \n",
        "#@markdown > Peaks indicate residues likely to be involved in beta-sheet \n",
        "#@markdown > aggregation.\n",
        "\n",
        "fig = plt.figure(figsize=plt.figaspect(0.5))\n",
        "ax1 = fig.add_subplot(1, 1, 1, projection='3d')\n",
        "\n",
        "fig.set_figwidth(15)\n",
        "fig.set_figheight(10)\n",
        "\n",
        "fig.suptitle(f'Single Sequence Predictions (3D Projection)')\n",
        "\n",
        "for index, seq_key in enumerate(keys):\n",
        "    predictions = all_predictions[seq_key]\n",
        "    x_position = range(len(predictions['seq']))\n",
        "    z_position = [index] * len(predictions['seq'])\n",
        "\n",
        "    agmata_pred = predictions['agmata']\n",
        "\n",
        "    ax1.plot(x_position, z_position, agmata_pred, label=seq_key)\n",
        "\n",
        "ax1.set_title('Agmata aggregation propensity')\n",
        "ax1.set_xlabel('residue index')\n",
        "ax1.set_zlabel('prediction values')\n",
        "ax1.set_yticks(range(len(keys)))\n",
        "ax1.set_yticklabels(keys)\n",
        "ax1.view_init(60, -45)\n",
        "\n",
        "plt.savefig('/content/results/agmata_predictons_3d.png')\n",
        "plt.show()"
      ],
      "metadata": {
        "cellView": "form",
        "id": "oRxE84upgFdn"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Download the results\n",
        "#@markdown After running this cell you will be able to download the plots\n",
        "#@markdown plain results in JSON format to your computer.\n",
        "\n",
        "json.dump(all_predictions, open('/content/results/b2btools_results.json', 'w'), indent=4, sort_keys=True)\n",
        "\n",
        "!zip -r /content/b2b-tools-results.zip /content/results\n",
        "\n",
        "files.download(\"/content/b2b-tools-results.zip\")"
      ],
      "metadata": {
        "cellView": "form",
        "id": "NyBFvJ4vMqBJ"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}