Upload dendrogram + PCA clustering script

scripts/tcga_fpkm_dendrogram/README.md

@@ -0,0 +1,53 @@
+# TCGA_FPKM_dendrogram
+
+Generates dendrogram image and csv file holding TCGA patients and their assigned cluster colors.
+
+Additional optional functionality:
+- Generates csv file holding TCGA patients clustered into n clusters using dendrogram's comparison function
+- Generate interactive dendrogram using Plotly library
+- Generate scatter plot of 2-component PCA analysis using TCGA-FPKM data
+
+## Installation
+
+`virtualenv` is a way to isolate software dependencies for python scripts.
+
+First install `virtualenv`:
+
+```bash
+pip3 install virtualenv
+```
+
+Then setup and activate the virtualenv:
+```bash
+cd tcga_fpkm_dendogram
+python3 -m venv env
+source env/bin/activate
+```
+
+Then install the dependencies for this script:
+
+```bash
+pip install wheel
+pip install -r requirements.txt
+```
+
+Now you are ready to run the script.
+
+IMPORTANT: When you are done using the script, make sure to deactivate the `virtualenv` by running the `deactivate` command.
+
+```bash
+# Deactivates virtualenv and stops isolating dependencies.
+deactivate
+```
+
+Otherwise your other python scripts may not work.
+
+## Usage
+
+Once you have activated the `virtualenv`, use the script as follows:
+
+```bash
+python3 tcga_fpkm_dendrogram.py <filename>
+```
+
+A folder called `results` will be created which will contain the following output files:

scripts/tcga_fpkm_dendrogram/gen_dendro.py

@@ -0,0 +1,96 @@
+from sklearn.cluster import AgglomerativeClustering
+from sklearn.decomposition import PCA
+import scipy.cluster.hierarchy as shc
+import plotly.figure_factory as ff
+import matplotlib.pyplot as plt
+import pandas as pd
+import numpy as np
+import shutil
+import sys
+import os
+
+# Create and save TCGA-FPKM static dendrogram & patient assigned cluster colors
+def plot_dendro_static(data, labels):
+    plt.figure(figsize=(10, 7))
+    plt.title("TCGA-FPKM: Dendrogram")
+    result = shc.dendrogram(shc.linkage(data, method='ward'))
+    plt.savefig('./results/TCGA-FPKM_dendrogram-stat.png')
+    plt.show()
+
+    leaf_colors = {}
+    color_tcga = {}
+    missed_tcga = patient_labels[845]
+
+    for i in range(len(result['color_list'])):
+        leaf_colors[result['leaves'][i]] = result['color_list'][i]
+
+    for j in range(len(patient_labels)):
+        if j != 845:
+            color_tcga[patient_labels[j]] = leaf_colors[j]
+
+    df = pd.DataFrame({'Patient': list(color_tcga.keys()), 
+                        'Cluster_color': list(color_tcga.values())})
+    df.to_csv('./results/TCGA-FPKM_cluster-colors.csv', index=False)
+
+# Create and save TCGA-FPKM interactive dendrogram
+def plot_dendro_interactive(data, labels):
+    data_array = tcga_data.to_numpy()
+    fig = ff.create_dendrogram(data_array, orientation='left', labels=labels)
+    fig.update_layout(width=1600, height=800)
+    fig.write_image('./results/TCGA-FPKM_dendrogram-intr.png')
+    fig.show()
+
+# Create and save TCGA-FPKM patient instances clustered into n clusters
+def gen_clusters(n, data, labels):
+    cluster = AgglomerativeClustering(n_clusters=n, affinity='euclidean', 
+                                        linkage='ward')
+    cluster.fit_predict(data)
+    clusters = cluster.labels_.tolist()
+
+    df = pd.DataFrame(list(zip(labels, clusters)), 
+                        columns =['Patient', 'Cluster'])
+    df.to_csv('./results/TCGA-FPKM_Cluster.csv', index=False)
+
+# Create and save 2-component PCA analysis scatter plot of TCGA-FPKM data
+def gen_scatter(data):
+        features = list(tcga_data.columns)
+        x = tcga_data.loc[:, features].values
+
+        pca = PCA(n_components=2)
+        principalComponents = pca.fit_transform(x)
+        principalDf = pd.DataFrame(data = principalComponents
+                , columns = ['principal component 1', 'principal component 2'])
+
+        principalDf.plot.scatter(x='principal component 1', 
+                                    y='principal component 2', 
+                                    title= "TCGA-FPKM: 2 Component PCA")
+        plt.savefig('./results/TCGA-FPKM_pca_scatter.png')
+        plt.show(block=True)
+
+#################################### MAIN #####################################
+if __name__ == "__main__":
+    # Check command-line argument
+    if (len(sys.argv) < 2):
+        print('missing input file name argument')
+        sys.exit()
+
+    # Check file validity
+    filename = str(sys.argv[1])
+    if (filename.split('.')[-1] != 'csv'):
+        print('input file is invalid type (not .csv)')
+        sys.exit()
+
+    # Create directory for results
+    path = './results'
+    if (os.path.isdir(path)):
+        shutil.rmtree(path)
+    os.mkdir(path)
+
+    # Reading in TCGA data
+    tcga_df = pd.read_csv(filename)
+    patient_labels = tcga_df.columns[1:]
+    tcga_data = tcga_df.drop(columns=['sample']).transpose()
+
+    # Plotting static dendrogram
+    plot_dendro_static(tcga_data, patient_labels)
+

scripts/tcga_fpkm_dendrogram/requirements.txt

@@ -0,0 +1,6 @@
+plotly==4.11.0
+scipy==1.4.1
+pandas==1.0.4
+numpy==1.18.5
+matplotlib==3.2.1
+scikit_learn==0.23.2