sample_scripts.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "# Sample scripts accompanying \"Comprehensive computational analysis of epigenetic descriptors affecting CRISPR-Cas9 off-target activity\""
      ],
      "metadata": {
        "id": "ITRhfXYh-A9y"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Setup and Installation"
      ],
      "metadata": {
        "id": "bg5mlfb-6gOY"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "DOfolk4TN-pS"
      },
      "outputs": [],
      "source": [
        "#@title Set Folder Prefix\n",
        "\n",
        "##########################################\n",
        "######## EDIT AND RUN THIS FIRST #########\n",
        "##########################################\n",
        "\n",
        "# Set this to the folder where the python scripts and IPython notebooks\n",
        "# are located\n",
        "folder_prefix = '/content/drive/MyDrive/nuc_occup/sample_script/'"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zkGj07FufZJF",
        "outputId": "f80af6bc-1b09-496a-8343-852afc724779"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Requirement already satisfied: xgboost in /usr/local/lib/python3.7/dist-packages (0.90)\n",
            "Collecting xgboost\n",
            "  Downloading xgboost-1.6.2-py3-none-manylinux2014_x86_64.whl (255.9 MB)\n",
            "\u001b[K     |████████████████████████████████| 255.9 MB 32 kB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from xgboost) (1.21.6)\n",
            "Requirement already satisfied: scipy in /usr/local/lib/python3.7/dist-packages (from xgboost) (1.7.3)\n",
            "Installing collected packages: xgboost\n",
            "  Attempting uninstall: xgboost\n",
            "    Found existing installation: xgboost 0.90\n",
            "    Uninstalling xgboost-0.90:\n",
            "      Successfully uninstalled xgboost-0.90\n",
            "Successfully installed xgboost-1.6.2\n"
          ]
        }
      ],
      "source": [
        "#@title Upgrade XGBoost\n",
        "!pip install xgboost --upgrade # the code works on xgboost >=1.6.2\n",
        "# RESTART THE RUNTIME IF XGBOOST HAS UPGRADED"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Sample Scripts for XGBoost model"
      ],
      "metadata": {
        "id": "6QFR8XLs7PKN"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### 1. Example Usage of XGBoost model\n",
        "Corresponds to xgb_predict_one.py"
      ],
      "metadata": {
        "id": "GAfibS_b6nP3"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cyDEGYtquPD2",
        "outputId": "986fb404-5b84-410e-aafa-efe6212a51a9"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Predicted CRISPR-Cas9 Cleavage Activity: -1.054\n"
          ]
        }
      ],
      "source": [
        "import pandas as pd\n",
        "import xgboost as xgb\n",
        "\n",
        "# This Python script uses the following data and model files:\n",
        "# 1. An example input CSV file with one row, namely one CRISPR-Cas9 cleavage\n",
        "#    activity datapoint from the crisprSQL dataset in the paper\n",
        "example_loc = folder_prefix + 'data/example_input.csv'\n",
        "# 2. The XGBoost tree model used for making CRISPR-Cas9 cleavage activity\n",
        "#    prediction\n",
        "xgb_loc = folder_prefix + 'models/xgb_model.json'\n",
        "\n",
        "# Create and load the XGBoost model\n",
        "# Use XGBRegressor since CRISPR-Cas9 cleavage activity values are continuous\n",
        "xgb_model = xgb.XGBRegressor()\n",
        "xgb_model.load_model(xgb_loc)\n",
        "\n",
        "# Read the input CSV file\n",
        "example_df = pd.read_csv(example_loc)\n",
        "\n",
        "# Make the prediction\n",
        "pred = xgb_model.predict(example_df.astype('float'))\n",
        "\n",
        "# Print the result\n",
        "# -1.054 is the predicted CRISPR-Cas9 cleavage activity value\n",
        "print('Predicted CRISPR-Cas9 Cleavage Activity: {:.3f}'.format(pred[0]))"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### 2. Use XGBoost model to predict CRISPR cleavage activity for 2000 randomly selected points from the  crisprSQL dataset\n",
        "Corresponds to xgb_predict.py"
      ],
      "metadata": {
        "id": "uKcbAdb96uGL"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9WKId3qNiCnr",
        "outputId": "6a3bacd9-8ab8-44cd-83d2-0c17ccbe1234"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Predicted CRISPR-Cas9 Cleavage Activities: [-1.0535808 -3.4915433 -1.7820611 ... -0.7105575 -1.2672865 -4.081657 ]\n",
            "Read the full list of predictions at /content/drive/MyDrive/nuc_occup/sample_script/out/xgb_preds.csv\n"
          ]
        }
      ],
      "source": [
        "import pandas as pd\n",
        "import xgboost as xgb\n",
        "\n",
        "# This Python script uses the following data and model files:\n",
        "# 1. An input CSV file containing 2000 random experimental/augmented\n",
        "#    CRISPR-Cas9 cleavage activity datapoints from the crisprSQL dataset\n",
        "#    in the paper\n",
        "csv_loc = folder_prefix + 'data/crisprSQL_dataset_2000.csv'\n",
        "# 2. The XGBoost tree model used for making CRISPR-Cas9 cleavage activity\n",
        "#    prediction\n",
        "xgb_loc = folder_prefix + 'models/xgb_model.json'\n",
        "\n",
        "# We will save the predictions in the \"out\" folder\n",
        "out_loc = folder_prefix + 'out/xgb_preds.csv'\n",
        "\n",
        "# Create and load the XGBoost model\n",
        "# Use XGBRegressor since CRISPR-Cas9 cleavage activity values are continuous\n",
        "xgb_model = xgb.XGBRegressor()\n",
        "xgb_model.load_model(xgb_loc)\n",
        "\n",
        "# Read the input CSV file\n",
        "X_df = pd.read_csv(csv_loc)\n",
        "\n",
        "# Make predictions\n",
        "preds = xgb_model.predict(X_df)\n",
        "\n",
        "# Save the predictions\n",
        "pd.DataFrame(preds).to_csv(out_loc, index=None)\n",
        "\n",
        "# Print results\n",
        "# Predicted CRISPR-Cas9 Cleavage Activities: [-1.0535808 -3.4915433 -1.7820611 ...\n",
        "# -0.7105575 -1.2672865 -4.081657 ]\n",
        "print('Predicted CRISPR-Cas9 Cleavage Activities: ' + str(preds))\n",
        "print('Read the full list of predictions at ' + out_loc)"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### 3. Varying Nucleotide BDM, GC147 and NuPoP (Affinity) values in XGBoost model\n",
        "Corresponds to xgb_vary.py"
      ],
      "metadata": {
        "id": "pidMdAys7MuE"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 861
        },
        "id": "fF37NVM0fWdt",
        "outputId": "878d8a71-1f42-4c3f-fdd9-4c7a970c96fa"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Nucleotide BDM's value ranges from 172.66 to 480.48\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "GC147's value ranges from 0.25 to 0.86\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "NuPoP (Affinity)'s value ranges from 0.50 to 2.13\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import xgboost as xgb\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# In the paper, we identified Nucleotide BDM, GC147 and NuPoP (Affinity) as\n",
        "# important computed nucleosome organization-related scores. Thus, in this\n",
        "# Python script, we vary the values of these scores, and see how such varied\n",
        "# values affect XGBoost's predictions.\n",
        "\n",
        "# This Python script uses the following data and model files:\n",
        "# 1. An input CSV file containing 2000 random experimental/augmented datapoints\n",
        "#    from the crisprSQL dataset in the paper\n",
        "csv_loc = folder_prefix + 'data/crisprSQL_dataset_2000.csv'\n",
        "# 2. The XGBoost tree model used for making CRISPR-Cas9 cleavage activity\n",
        "#    prediction\n",
        "xgb_loc = folder_prefix + 'models/xgb_model.json'\n",
        "\n",
        "# The plots generated will be saved in these relative locations\n",
        "out_dict = {'Nucleotide BDM': 'out/xgb_vary_nucleotide_bdm.pdf',\n",
        "            'GC147': 'out/xgb_vary_gc147.pdf',\n",
        "            'NuPoP (Affinity)': 'out/xgb_vary_nupop_affinity.pdf'}\n",
        "\n",
        "### IMPORTANT CONSTANTS ### \n",
        "# Don't change these!\n",
        "TAR_SEQ_LEN = 23 # target DNA sequence's length\n",
        "\n",
        "# Feel free to change these\n",
        "NUM_SAMPLES = 5 # Number of random datapoints to select for visualization,\n",
        "                 # i.e., number of row to select from the input CSV file\n",
        "NUM_VARIED = 20 # Number of varied points per random datapoint\n",
        "###########################\n",
        "\n",
        "# Read the input CSV file\n",
        "X_df = pd.read_csv(csv_loc)\n",
        "\n",
        "# Create and load the XGBoost model\n",
        "# Use XGBRegressor since CRISPR-Cas9 cleavage activity values are continuous\n",
        "xgb_model = xgb.XGBRegressor()\n",
        "xgb_model.load_model(xgb_loc)\n",
        "\n",
        "# These are the column/feature names corresponding to ...\n",
        "# (a) Nucleotide BDM\n",
        "NBDM_feats = ['NucleotideBDM_' + str(i) for i in range(1, 1 + TAR_SEQ_LEN)]\n",
        "# (b) GC147\n",
        "GC_feats = ['GCContent_' + str(i) for i in range(1, 1 + TAR_SEQ_LEN)]\n",
        "# (c) NuPoP (Affinity)\n",
        "NuPoP_Aff_feats = ['NuPoP_Affinity_147_h'] + \\\n",
        "            ['NuPoP_Affinity_147_h.' + str(i) for i in range(1, TAR_SEQ_LEN)]\n",
        "# Put these features into a dictionary\n",
        "feats_dict = {'Nucleotide BDM': NBDM_feats,\n",
        "              'GC147': GC_feats,\n",
        "              'NuPoP (Affinity)': NuPoP_Aff_feats}\n",
        "\n",
        "# To obtain sensible value ranges, we exclude data points with zeros across\n",
        "# all computed nucleosome organization-related score, since these zeros arise\n",
        "# from data imputation during training.\n",
        "non_zero_X_df = X_df[X_df['GCContent_1'] > 0]\n",
        "\n",
        "# Set values ranges for each computed score to be within the minimum and\n",
        "# maximum value of the respective features, i.e., we vary within the ranges\n",
        "# of the dataset.\n",
        "ranges_dict = {'Nucleotide BDM': (non_zero_X_df[NBDM_feats].min(axis=1).min(),\n",
        "                                  non_zero_X_df[NBDM_feats].max(axis=1).max()),\n",
        "              'GC147': (non_zero_X_df[GC_feats].min(axis=1).min(),\n",
        "                        non_zero_X_df[GC_feats].max(axis=1).max()),\n",
        "              'NuPoP (Affinity)': (non_zero_X_df[NuPoP_Aff_feats].min(axis=1).min(),\n",
        "                                   non_zero_X_df[NuPoP_Aff_feats].max(axis=1).max())}\n",
        "\n",
        "# We select NUM_SAMPLES random data points...\n",
        "X_sample = X_df.sample(NUM_SAMPLES)\n",
        "\n",
        "# Generate plots for each of the three computed scores\n",
        "for feat_name, feats in feats_dict.items():\n",
        "  min_val, max_val = ranges_dict[feat_name]\n",
        "  print(\"{}'s value ranges from {:.2f} to {:.2f}\".format(feat_name,\n",
        "                                                         min_val, max_val))\n",
        "\n",
        "  # For each data point...\n",
        "  for i in range(NUM_SAMPLES):\n",
        "    point_chosen = X_sample.iloc[i]\n",
        "\n",
        "    # Create a Pandas dataframe for the random datapoint\n",
        "    example_df = pd.DataFrame(point_chosen).transpose()\n",
        "\n",
        "    # Determine the varied values to change to\n",
        "    varied_vals = np.linspace(min_val, max_val, NUM_VARIED)\n",
        "\n",
        "    # Create the artificial datapoints containing the varied values\n",
        "    vary_df = example_df.loc[example_df.index.repeat(len(varied_vals))]\n",
        "    for feat in feats:\n",
        "      vary_df[feat] = varied_vals\n",
        "\n",
        "    # Make predictions for these artificial datapoints\n",
        "    preds = xgb_model.predict(vary_df)\n",
        "\n",
        "    # Plot the line\n",
        "    plt.plot(varied_vals, preds)\n",
        "\n",
        "  # Set...\n",
        "  # the plot's minimum and maximum x-value, ...\n",
        "  plt.xlim(min_val, max_val)\n",
        "  # the x-axis's name, and\n",
        "  plt.xlabel(\"Varied {} Value\".format(feat_name))\n",
        "  # the y-axis's name\n",
        "  plt.ylabel(\"Predicted CRISPR-Cas9 Cleavage Activity\")\n",
        "\n",
        "  # Save the plot\n",
        "  plt.savefig(folder_prefix + out_dict[feat_name])\n",
        "\n",
        "  # This is a Jupyter notebook, so let's show the figures when running\n",
        "  # this cell\n",
        "  plt.show()\n",
        "\n",
        "  # Close the current figure\n",
        "  plt.close()\n",
        " \n",
        "# The script prints:\n",
        "# Nucleotide BDM's value ranges from 172.66 to 480.48\n",
        "# GC147's value ranges from 0.25 to 0.86\n",
        "# NuPoP (Affinity)'s value ranges from 0.50 to 2.13\n",
        "\n",
        "# We observe the following tendencies in the plots:\n",
        "#   - low Nucleotide BDM value corresponds to low predicted CRISPR-Cas9 cleavage\n",
        "#     activity value\n",
        "#   - increasing GC147 increases predicted CRISPR-Cas9 cleavage activity value\n",
        "#   - high NuPoP (Affinity) value corresponds to low predicted CRISPR-Cas9\n",
        "#     cleavage activity value"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XqsmM7l-5FXv"
      },
      "source": [
        "## Sample Scripts for convolutional neural network (CNN) model"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### 0. PyTorch CNN model architecture\n",
        "\n",
        "RUN THIS BEFORE ANY OF THE CODE BELOW\n",
        "\n",
        "Corresponds to cnn.py"
      ],
      "metadata": {
        "id": "UPrp14CRAnTb"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "X2lkWjcj5OrA"
      },
      "outputs": [],
      "source": [
        "import torch\n",
        "import pickle\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "\n",
        "### IMPORTANT CONSTANTS ###\n",
        "tarSeqLen = 23 # target DNA sequence's length\n",
        "###########################\n",
        "\n",
        "class GaussianNoise(torch.nn.Module):\n",
        "  ''' Custom module to implement Gaussian random noise\n",
        "  in nn.Sequential containers.\n",
        "\n",
        "  Attributes:\n",
        "    mean: Float indicating the Gaussian noise's mean.\n",
        "    stdev: Float indicating the Gaussian noise's standard deviation.\n",
        "  '''\n",
        "  def __init__(self, mean, stdev):\n",
        "    ''' Initializes GaussianNoise with mean and standard deviation.\n",
        "    Input:\n",
        "      mean: Float indicating the Gaussian noise's mean.\n",
        "      stdev: Float indicating the Gaussian noise's standard deviation.\n",
        "    Output:\n",
        "      None\n",
        "    '''\n",
        "    super(GaussianNoise, self).__init__()\n",
        "    self.mean = mean\n",
        "    self.stdev = stdev\n",
        "\n",
        "  def forward(self, ins):\n",
        "    ''' Forward pass in layer. Only adds noise during training. '''\n",
        "    if self.training:\n",
        "      noise = ins.data.new(ins.size()).normal_(self.mean, self.stdev)\n",
        "      return ins + noise\n",
        "    return ins\n",
        "\n",
        "class ConvolutionalNet(torch.nn.Module):\n",
        "  ''' PyTorch convolutional neural network (CNN) regression model used for\n",
        "  predicting CRISPR-Cas9 (off-)target activity values.\n",
        "\n",
        "  Non-layer attributes:\n",
        "    device: The compute device used for tensor computations (e.g., cpu, gpu).\n",
        "    p: Dropout probability used in EncodeLayer1's nn.Dropout layer.\n",
        "    mean: Mean used in EncodeLayer1's GaussianNoise layer.\n",
        "    stdev: Standard deviation used in EncodeLayer1's GaussianNoise layer.\n",
        "    batchnorm_momentum: Momentum of BatchNorm's running estimates.\n",
        "    epiDim: Number of epigenetic features (always 22).\n",
        "  \n",
        "  Layer attributes:\n",
        "    BatchNorm: 1D BatchNorm layer.\n",
        "    EncodeLayer1: nn.Sequential block consisting of a Conv1d,\n",
        "                  GaussianNoise, Dropout and then a LeakyReLU layer.\n",
        "    EncodeLayer2: nn.Sequential block consisting of a Conv1d\n",
        "                  and then a LeakyReLU.\n",
        "    EncodeLayer3: nn.Sequential block consisting of a Conv1d, BatchNorm\n",
        "                  and then a LeakyReLU.\n",
        "    ConjoinedLayer1: nn.Sequential block consisting of a Conv1d, MaxPool1d\n",
        "                  and then a ReLU.\n",
        "    ConjoinedLayer2: nn.Sequential block consisting of a Conv1d, MaxPool1d\n",
        "                  and then a ReLU.\n",
        "    conjoinedLinear: nn.Linear layer.\n",
        "  '''\n",
        "  \n",
        "  def __init__(self, epiDim, device=None,\n",
        "               p = 0.0, mean = 0, stdev = 0.254, batchnorm_momentum = 0.1):\n",
        "    ''' Initializes ConvolutionalNet with training parameters\n",
        "    and neural net layers\n",
        "\n",
        "    Input:\n",
        "      epiDim: Number of epigenetic features (always 22).\n",
        "      device: The compute device used for tensor computations\n",
        "              (e.g., cpu, gpu).\n",
        "      p: Dropout probability used in EncodeLayer1's nn.Dropout layer.\n",
        "      mean: Mean used in EncodeLayer1's GaussianNoise layer.\n",
        "      stdev: Standard deviation used in EncodeLayer1's GaussianNoise layer.\n",
        "      batchnorm_momentum: Momentum of BatchNorm's running estimates.\n",
        "    \n",
        "    Output:\n",
        "      None\n",
        "    '''\n",
        "    super(ConvolutionalNet, self).__init__()\n",
        "    self.device = device\n",
        "    self.p, self.mean, self.stdev = p, mean, stdev\n",
        "    self.batchnorm_momentum = batchnorm_momentum\n",
        "    self.epiDim = epiDim\n",
        "    self.BatchNorm = torch.nn.BatchNorm1d(self.epiDim,\n",
        "                                          momentum=self.batchnorm_momentum)\n",
        "    self.BatchNorm.to(self.device)\n",
        "\n",
        "    # Encode Layers\n",
        "    # don't set batchnorm_momentum as this will probably mess up\n",
        "    # the running stats\n",
        "    self.EncodeLayer1 = torch.nn.Sequential(\n",
        "      torch.nn.Conv1d(int(self.epiDim), 32, kernel_size=3, stride=2,\n",
        "                      padding=0),\n",
        "      GaussianNoise(self.mean, self.stdev), # set Gaussian mean and stdev\n",
        "      torch.nn.Dropout(p=self.p), # set dropout p\n",
        "      torch.nn.LeakyReLU(0.2)\n",
        "      )\n",
        "    self.EncodeLayer1.to(self.device)\n",
        "\n",
        "    self.EncodeLayer2 = torch.nn.Sequential(\n",
        "      torch.nn.Conv1d(32, 64, kernel_size=3, stride=1, padding=0),\n",
        "      torch.nn.LeakyReLU(0.2)\n",
        "      )\n",
        "    self.EncodeLayer2.to(self.device)\n",
        "\n",
        "\n",
        "    self.EncodeLayer3 = torch.nn.Sequential(\n",
        "      torch.nn.Conv1d(64, 128, kernel_size=3, stride=2, padding=0),\n",
        "      torch.nn.BatchNorm1d(128),\n",
        "      torch.nn.LeakyReLU(0.2)\n",
        "      )\n",
        "    self.EncodeLayer3.to(self.device)\n",
        "\n",
        "    # Conjoined Layers\n",
        "    self.conjoinedLayer1 = torch.nn.Sequential(\n",
        "      torch.nn.Conv1d(128, 256, kernel_size=3, stride=1, padding=0),\n",
        "      torch.nn.MaxPool1d(kernel_size=3, padding=1, stride=1), # preserve\n",
        "                                                              # dimensions\n",
        "      torch.nn.ReLU()\n",
        "      )\n",
        "    self.conjoinedLayer1.to(self.device)\n",
        "\n",
        "    self.conjoinedLayer2 = torch.nn.Sequential(\n",
        "      torch.nn.Conv1d(256, 512, kernel_size=2, stride=1, padding=0),\n",
        "      torch.nn.MaxPool1d(kernel_size=3, padding=1, stride=1),\n",
        "      torch.nn.ReLU()\n",
        "      )\n",
        "    self.conjoinedLayer2.to(self.device)\n",
        "\n",
        "    self.conjoinedLinear = torch.nn.Linear(512, 1)\n",
        "    self.conjoinedLinear.to(self.device)\n",
        "\n",
        "  def forward(self, x):\n",
        "    ''' Performs a forward pass using the instantiated ConvolutionalNet\n",
        "    model.\n",
        "\n",
        "    Input:\n",
        "      x: PyTorch tensor as input to the ConvolutionalNet model.\n",
        "         The tensor contains the 22 epigenetic features.\n",
        "         Dimension: (X.shape[0], epiDim, tarSeqLen), where\n",
        "           - X.shape[0]         is the # of datapoints,\n",
        "           - epiDim (== 22)     is the # of epigenetic features \n",
        "           - tarSeqLen (== 23)  is the target sequence's length.\n",
        "\n",
        "    Output:\n",
        "      out: PyTorch tensor containing predicted CRISPR-Cas9 (off-)target\n",
        "           cleavage activity values.\n",
        "    '''\n",
        "    x = x.view(x.size(0), self.epiDim, -1).to(self.device)\n",
        "    x = self.BatchNorm(x)\n",
        "\n",
        "    # forward pass thru encode layers\n",
        "    x_enc = self.EncodeLayer1(x)\n",
        "    x_enc = self.EncodeLayer2(x_enc)\n",
        "    seq_encoding = self.EncodeLayer3(x_enc)\n",
        "\n",
        "    # forward pass thru conjoined layers\n",
        "    out = self.conjoinedLayer1(seq_encoding)\n",
        "    out = self.conjoinedLayer2(out)\n",
        "\n",
        "    # flatten last axis to 512 x 1\n",
        "    out = out.reshape(out.size(0), -1)\n",
        "    \n",
        "    # forward pass thru linear layer\n",
        "    out = self.conjoinedLinear(out)\n",
        "    return out # regression\n",
        "\n",
        "def vecToMatEncoding(X, numBpWise=0):\n",
        "  ''' Formats the epigenetic feature-containing PyTorch tensor so that it\n",
        "      is ready as input to the ConvolutionalNet model.\n",
        "  \n",
        "      To do this, we first transpose the first numBpWise*tarSeqLen columns\n",
        "      in PyTorch Tensor X. These columns come in sets of tarSeqLen, where\n",
        "      each set corresponds to a base pair-resolved computed nucleosome\n",
        "      organization score. Then, for the remaining columns in X, we repeat\n",
        "      each column tarSeqLen times. Finally, we concatenate the two tensors\n",
        "      together to form the output. Here is a conceptual visualization of how\n",
        "      this function transforms each column in X\n",
        "\n",
        "      bp-resolved column features: a, b, ..., m (there are 13 of these)\n",
        "      non-bp-resolved column features: n, o,..., v (there are 9 of these)\n",
        "                | vecToMatEncoding\n",
        "                |\n",
        "                v\n",
        "      X (size 308)     : [aa...abb...b...mm...mno...v]\n",
        "      output (size 22 x 23): [[a...a], [b...b], ..., [m...m],\n",
        "                              [n...n], [o...o], ..., [v...v],]\n",
        "\n",
        "  Input:\n",
        "    X: 2D PyTorch tensor. Each row represents a datapoint, and\n",
        "       each column represents an epigenetic feature\n",
        "       Dimension: (# of datapoints, 308), where\n",
        "       308 = numBpWise * tarSeqLen\n",
        "             + (# of non-bp-resolved (energy/experimental) epigenetic scores)\n",
        "           = 13 * 23 + 9\n",
        "    numBpWise: Number of base pair-resolved nucleosome organization-related\n",
        "               scores/features (always 13).\n",
        "  \n",
        "  Output:\n",
        "    PyTorch tensor ready for forward pass.\n",
        "    Dimension: (X.shape[0], epiDim, tarSeqLen), where\n",
        "      - X.shape[0]         is the # of datapoints,\n",
        "      - epiDim (== 22)     is the # of epigenetic features \n",
        "      - tarSeqLen (== 23)  is the target sequence's length.\n",
        "  '''\n",
        "  \n",
        "  bpwise_features = []\n",
        "  for i in range(numBpWise):\n",
        "    # we format each base pair-wise feature to have\n",
        "    # dimension (X.shape[0], 1, tarSeqLen)\n",
        "    bpwise_features.append(X[:, i*tarSeqLen:(i+1)*tarSeqLen] \\\n",
        "                    .view(X.shape[0], tarSeqLen, 1) \\\n",
        "                    .transpose(1,2))\n",
        "\n",
        "  # None indexing means adding an axis - could also use .unsqueeze() here\n",
        "  epigenetics = X[:, numBpWise*tarSeqLen:, None].repeat(1, 1, tarSeqLen)\n",
        "\n",
        "  # return epigenetics and bpwise_features\n",
        "  return torch.cat(tuple(bpwise_features)+(epigenetics,), dim=1)"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### 1. Example Usage of the PyTorch CNN model\n",
        "Corresponds to cnn_predict_one.py"
      ],
      "metadata": {
        "id": "jKNfJUKg8oqB"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import pandas as pd\n",
        "import torch\n",
        "\n",
        "# This Python script uses the following data and model files:\n",
        "# 1. An example input CSV file with one row, namely one datapoint\n",
        "# from the crisprSQL dataset in the paper\n",
        "example_loc = folder_prefix + 'data/example_input.csv'\n",
        "# 2. PyTorch model file containing the convolutional neural network (CNN) model\n",
        "#    which predicts CRISPR-Cas9 (off-)target cleavage activity.\n",
        "state_dict_loc = folder_prefix + 'models/cnn_model.pt'\n",
        "\n",
        "### IMPORTANT CONSTANTS ###\n",
        "numBpWise = 13 # Number of base pair-resolved computed\n",
        "               # nucleosome organization-related scores/features\n",
        "epiDim = 22 # The total number of epigenetic features considered\n",
        "###########################\n",
        "\n",
        "# Read in the crisprSQL data points\n",
        "X_df = pd.read_csv(example_loc)\n",
        "\n",
        "# Convert the Pandas DataFrame into a PyTorch tensor\n",
        "X_torch = torch.as_tensor(X_df.to_numpy()).float()\n",
        "\n",
        "# Decide which compute device to use when using PyTorch.\n",
        "# Use a GPU if available, CPU otherwise.\n",
        "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
        "\n",
        "# Create convolutional neural network regression PyTorch model\n",
        "torch_model = ConvolutionalNet(epiDim, device=device)\n",
        "\n",
        "# Load saved PyTorch weights\n",
        "state_dict = torch.load(state_dict_loc, map_location=device)\n",
        "torch_model.load_state_dict(state_dict)\n",
        "\n",
        "# Format the PyTorch tensor to fit the CNN's input dimensions\n",
        "x = vecToMatEncoding(X_torch, numBpWise=numBpWise)\n",
        "\n",
        "# Prepare for model inference by setting torch_model to evaluation mode \n",
        "torch_model.eval()\n",
        "\n",
        "# Predict the CRISPR-Cas9 cleavage activity\n",
        "pred = torch_model(x).cpu().flatten().detach().numpy()\n",
        "\n",
        "# Print results\n",
        "print('Predicted CRISPR-Cas9 Cleavage Activity: {:.3f}'.format(pred[0]))\n",
        "# on CPU/GPU: -0.669"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4_2ikSJnFgwE",
        "outputId": "9b9c874f-8653-4c98-aa10-6faca3a6831f"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Predicted CRISPR-Cas9 Cleavage Activity: -0.669\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### 2. Use PyTorch CNN model to predict CRISPR activity for 2000 randomly selected points from the crisprSQL dataset\n",
        "Corresponds to cnn_predict.py"
      ],
      "metadata": {
        "id": "m3yHSAuD80xb"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jmuGz9wI5zJa",
        "outputId": "2b2a0fb1-acd1-44e8-9196-78dc2ba4f45a"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Predicted CRISPR-Cas9 Cleavage Activities: [-0.669225   -4.1682196  -1.7856159  ... -0.58595705 -0.9737004\n",
            " -4.1463737 ]\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import torch\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# This Python script uses the following data and model files:\n",
        "# 1. An input CSV file containing 2000 random experimental/augmented datapoints\n",
        "#    from the crisprSQL dataset in the paper\n",
        "input_feat_loc = folder_prefix + 'data/crisprSQL_dataset_2000.csv'\n",
        "# 2. PyTorch model file containing the convolutional neural network (CNN) model\n",
        "#    which predicts CRISPR-Cas9 (off-)target cleavage activity.\n",
        "state_dict_loc = folder_prefix + 'models/cnn_model.pt'\n",
        "\n",
        "# We will save the predictions in the \"out\" folder\n",
        "out_loc = folder_prefix + 'out/cnn_preds.csv'\n",
        "\n",
        "### IMPORTANT CONSTANTS ###\n",
        "numBpWise = 13 # Number of base pair-resolved computed\n",
        "               # nucleosome organization-related scores/features\n",
        "epiDim = 22 # The total number of epigenetic features considered\n",
        "###########################\n",
        "\n",
        "# Read in the crisprSQL data points\n",
        "X_df = pd.read_csv(input_feat_loc)\n",
        "\n",
        "# Convert the Pandas DataFrame into a PyTorch tensor\n",
        "X_torch = torch.as_tensor(X_df.to_numpy()).float()\n",
        "\n",
        "# Decide which compute device to use when using PyTorch.\n",
        "# Use a GPU if available, CPU otherwise.\n",
        "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
        "\n",
        "# Create convolutional neural network regression PyTorch model\n",
        "torch_model = ConvolutionalNet(epiDim, device=device)\n",
        "\n",
        "# Load saved PyTorch weights\n",
        "state_dict = torch.load(state_dict_loc, map_location=device)\n",
        "torch_model.load_state_dict(state_dict)\n",
        "\n",
        "# Format the PyTorch tensor to fit the CNN's input dimensions\n",
        "x = vecToMatEncoding(X_torch, numBpWise=numBpWise)\n",
        "\n",
        "# Prepare for model inference by setting torch_model to evaluation mode \n",
        "torch_model.eval()\n",
        "\n",
        "# Predict CRISPR-Cas9 cleavage activities\n",
        "preds_y = torch_model(x).cpu().flatten().detach().numpy()\n",
        "\n",
        "# Save the predictions\n",
        "pd.DataFrame(preds_y).to_csv(out_loc)\n",
        "\n",
        "# Print results\n",
        "print('Predicted CRISPR-Cas9 Cleavage Activities:', preds_y)\n",
        "# Result varies due to floating point precision arithmetic\n",
        "# Example outputs:\n",
        "# on CPU: [-0.669225   -4.1682196  -1.7856159  ... -0.58595717 -0.97370046\n",
        "# -4.1463737 ]\n",
        "# on GPU: [-0.6692251  -4.1682186  -1.7856162  ... -0.58595747 -0.9737008\n",
        "# -4.1463733 ]"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### 3. Varying Nucleotide BDM, GC147 and NuPoP (Affinity) values in PyTorch CNN model\n",
        "Corresponds to cnn_vary.py"
      ],
      "metadata": {
        "id": "eHlg1zFK84vP"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 858
        },
        "id": "mL-UeiSt0XpU",
        "outputId": "75b6a443-57ea-4ff8-98be-b6eb5708c640"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Nucleotide BDM's value ranges from 172.66 to 480.48\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "GC147's value ranges from 0.25 to 0.86\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "NuPoP (Affinity)'s value ranges from 0.50 to 2.13\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import torch\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# Nucleotide BDM, GC147 and NuPoP (Affinity) are identified as important\n",
        "# computed nucleosome organization-related scores. Thus, in this Python script,\n",
        "# we vary the values of these scores, and see how such varied values\n",
        "# affect XGBoost's predictions.\n",
        "\n",
        "# This Python script uses the following data and model files:\n",
        "# 1. An input CSV file containing 2000 random experimental/augmented datapoints\n",
        "#    from the crisprSQL dataset in the paper\n",
        "csv_loc = folder_prefix + 'data/crisprSQL_dataset_2000.csv'\n",
        "# 2. PyTorch model file containing the convolutional neural network (CNN) model\n",
        "#    which predicts CRISPR-Cas9 (off-)target cleavage activity.\n",
        "state_dict_loc = folder_prefix + 'models/cnn_model.pt'\n",
        "\n",
        "# The plots generated will be saved in these relative locations\n",
        "out_dict = {'Nucleotide BDM': 'out/cnn_vary_nucleotide_bdm.pdf',\n",
        "            'GC147': 'out/cnn_vary_gc147.pdf',\n",
        "            'NuPoP (Affinity)': 'out/cnn_vary_nupop_affinity.pdf'}\n",
        "\n",
        "### IMPORTANT CONSTANTS ### \n",
        "# Don't change these!\n",
        "TAR_SEQ_LEN = 23 # target DNA sequence's length\n",
        "epiDim = 22 # The total number of epigenetic features considered\n",
        "numBpWise = 13 # Number of base pair-resolved nucleosome organization-related\n",
        "               # scores/features\n",
        "\n",
        "# Feel free to change\n",
        "NUM_SAMPLES = 5 # Number of random datapoints to select for visualization,\n",
        "                 # i.e., number of row to select from the input CSV file\n",
        "NUM_VARIED = 20 # Number of varied points per random datapoint\n",
        "###########################\n",
        "\n",
        "# Read the input CSV file\n",
        "X_df = pd.read_csv(csv_loc)\n",
        "\n",
        "# Decide which compute device to use when using PyTorch.\n",
        "# Use a GPU if available, CPU otherwise.\n",
        "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
        "\n",
        "# Create convolutional neural network regression PyTorch model\n",
        "torch_model = ConvolutionalNet(epiDim, device=device)\n",
        "\n",
        "# Load saved PyTorch weights\n",
        "state_dict = torch.load(state_dict_loc, map_location=device)\n",
        "torch_model.load_state_dict(state_dict)\n",
        "\n",
        "# These are the column/feature names corresponding to ...\n",
        "# (a) Nucleotide BDM\n",
        "NBDM_feats = ['NucleotideBDM_' + str(i) for i in range(1, 1 + TAR_SEQ_LEN)]\n",
        "# (b) GC147\n",
        "GC_feats = ['GCContent_' + str(i) for i in range(1, 1 + TAR_SEQ_LEN)]\n",
        "# (c) NuPoP (Affinity)\n",
        "NuPoP_Aff_feats = ['NuPoP_Affinity_147_h'] + \\\n",
        "                  ['NuPoP_Affinity_147_h.' + str(i) for i in range(1,\n",
        "                                                                  TAR_SEQ_LEN)]\n",
        "# Put these features into a dictionary\n",
        "feats_dict = {'Nucleotide BDM': NBDM_feats,\n",
        "              'GC147': GC_feats,\n",
        "              'NuPoP (Affinity)': NuPoP_Aff_feats}\n",
        "\n",
        "# To obtain sensible value ranges, we exclude data points with zeros across\n",
        "# all computed nucleosome organization-related score, since these zeros arise\n",
        "# from data imputation during training.\n",
        "non_zero_X_df = X_df[X_df['GCContent_1'] > 0]\n",
        "\n",
        "# Set values ranges for each computed score to be within the minimum and\n",
        "# maximum value of the respective features, i.e., we vary within the ranges\n",
        "# of the dataset.\n",
        "ranges_dict = {'Nucleotide BDM': (non_zero_X_df[NBDM_feats].min(axis=1).min(),\n",
        "                                  non_zero_X_df[NBDM_feats].max(axis=1).max()),\n",
        "              'GC147': (non_zero_X_df[GC_feats].min(axis=1).min(),\n",
        "                        non_zero_X_df[GC_feats].max(axis=1).max()),\n",
        "              'NuPoP (Affinity)': (non_zero_X_df[NuPoP_Aff_feats].min(axis=1).min(),\n",
        "                                   non_zero_X_df[NuPoP_Aff_feats].max(axis=1).max())}\n",
        "\n",
        "# We select five random data points...\n",
        "X_sample = X_df.sample(NUM_SAMPLES)\n",
        "\n",
        "# Generate plots for each of the three computed scores\n",
        "for feat_name, feats in feats_dict.items():\n",
        "  min_val, max_val = ranges_dict[feat_name]\n",
        "  print(\"{}'s value ranges from {:.2f} to {:.2f}\".format(feat_name,\n",
        "                                                         min_val, max_val))\n",
        "\n",
        "  # For each data point...\n",
        "  for i in range(NUM_SAMPLES):\n",
        "    point_chosen = X_sample.iloc[i]\n",
        "\n",
        "    # Create a Pandas dataframe for the random datapoint\n",
        "    example_df = pd.DataFrame(point_chosen).transpose()\n",
        "\n",
        "    # Determine the varied values to change to\n",
        "    varied_vals = np.linspace(min_val, max_val, NUM_VARIED)\n",
        "\n",
        "    # Create the artificial datapoints containing the varied values\n",
        "    vary_df = example_df.loc[example_df.index.repeat(len(varied_vals))]\n",
        "    for feat in feats:\n",
        "      vary_df[feat] = varied_vals\n",
        "\n",
        "    # # Make predictions for these artificial datapoints\n",
        "    # preds = xgb_model.predict(vary_df)\n",
        "\n",
        "    # Convert the Pandas DataFrame into a PyTorch tensor\n",
        "    X_torch = torch.as_tensor(vary_df.to_numpy()).float()\n",
        "\n",
        "    # Format the PyTorch tensor to fit the CNN's input dimensions\n",
        "    x = vecToMatEncoding(X_torch, numBpWise=13)\n",
        "\n",
        "    # Prepare for model inference by setting torch_model to evaluation mode \n",
        "    torch_model.eval()\n",
        "\n",
        "    # Make CRISPR-Cas9 cleavage activity predictions for these artificial datapoints\n",
        "    preds = torch_model(x).cpu().flatten().detach().numpy()\n",
        "\n",
        "    # Plot the line\n",
        "    plt.plot(varied_vals, preds)\n",
        "\n",
        "  # Set...\n",
        "  # the plot's minimum and maximum x-value, ...\n",
        "  plt.xlim(min_val, max_val)\n",
        "  # the x-axis's name, and\n",
        "  plt.xlabel(\"Varied {} Value\".format(feat_name))\n",
        "  # the y-axis's name\n",
        "  plt.ylabel(\"Predicted CRISPR-Cas9 Cleavage Activity\")\n",
        "\n",
        "  # Save the plot\n",
        "  plt.savefig(folder_prefix + out_dict[feat_name])\n",
        "\n",
        "  # This is a Jupyter notebook, so let's show the figures when running\n",
        "  # this cell\n",
        "  plt.show()\n",
        "\n",
        "  # Close the current figure\n",
        "  plt.close()\n",
        " \n",
        "# The script prints:\n",
        "# Nucleotide BDM's value ranges from 172.66 to 480.48\n",
        "# GC147's value ranges from 0.25 to 0.86\n",
        "# NuPoP (Affinity)'s value ranges from 0.50 to 2.13\n",
        "\n",
        "# We observe the following tendencies in the plots:\n",
        "#   - increasing Nucleotide BDM increases predicted CRISPR-Cas9 cleavage\n",
        "#     activity value\n",
        "#   - increasing GC147 increases predicted CRISPR-Cas9 cleavage activity value\n",
        "#   - increasing NuPoP (Affinity) decreases predicted CRISPR-Cas9 cleavage\n",
        "#     activity value"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "collapsed_sections": [],
      "provenance": [],
      "toc_visible": true
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}