Added in rank order anaylsis, and analysis for cluster 3.

GRO_seq_TFSEE/cluster_3_tfs.csv

+symbol,ES_D0,ES_D2,ES_D5,ES_D7,ES_D10
+ATF4,212.425,195.03,196.87,163.705,170.855
+BACH1,10.78,10.375,12.685,15.295,11.275
+CENPB,20.49,15.53,20.99,21.09,25.775
+CREB1,33.125,28.415,30.6,30.75,29.89
+CTCF,31.91,26.56,32.98,25.825,35.205
+E2F4,45.005,31.09,42.27,33.1,39.23
+E2F6,12.765,12.05,13.85,15.7,13.92
+E2F8,3.76,0.915,4.715,0.98,5.53
+ELK4,6.175,5.23,7.985,4.75,5.61
+FOXD3,12.855,0.57,0.0,0.0,0.0
+FOXJ3,19.505,21.48,34.235,47.38,24.415
+FOXK1,2.245,7.215,8.445,6.16,7.32
+GABPA,7.11,9.695,9.135,7.24,7.335
+GLI2,17.815,4.495,1.7,0.37,1.075
+HINFP,17.6,17.74,20.445,22.71,16.795
+IRF2,6.345,5.31,6.055,8.94,6.925
+KLF1,0.32,0.805,1.18,0.59,0.45
+KLF16,17.695,18.915,10.375,6.245,6.515
+MAFG,14.79,14.55,9.585,13.625,12.465
+NR2C2,11.36,11.285,12.59,10.985,11.54
+NRL,0.635,1.2,0.86,0.67,1.13
+OTX2,16.925,441.385,30.42,3.66,1.185
+PLAG1,3.42,3.9,4.71,6.5,4.775
+RARG,13.71,12.405,9.065,4.305,2.855
+REST,14.235,18.47,22.54,21.62,17.925
+SP1,25.235,42.015,35.205,28.575,32.855
+SP2,6.75,6.92,10.8,8.075,12.81
+SP4,3.915,1.865,3.535,3.225,3.745
+SP8,1.075,2.045,0.515,0.295,0.05
+TCF4,21.4,12.385,19.12,26.95,17.01
+TFAP4,17.345,11.925,23.92,8.635,8.93
+ZIC3,27.06,95.46,4.98,0.695,0.085
+ZNF143,24.635,31.6,23.82,23.235,25.555
+ZNF263,11.53,10.735,14.55,12.815,15.215

GRO_seq_TFSEE/matrix_analysis.py

@@ -302,8 +302,8 @@ for h in headers:
 
 # Note 5 TFs not represented ['TCFE2A', 'RAR', 'ZFP423', 'RXR', 'TCFCP2L1']
 
-tf_cell_lines.columns = new_headers
-tf_cell_lines = tf_fpkm[subset]
+    tf_cell_lines.columns = new_headers
+    tf_cell_lines = tf_fpkm[subset]
 
 # Log2 scale FPKM
 x = tf_cell_lines.stack()
@@ -393,7 +393,7 @@ from sklearn.cluster import KMeans
 from sklearn.decomposition import PCA
 from sklearn.preprocessing import scale
 from sklearn.manifold import TSNE
-plt..style.use('classic')
+plt.style.use('classic')
 
 
 #TNSE
@@ -663,6 +663,81 @@ scipy.stats.ranksums(cluster4_enhancers['ES_D5'],cluster4_enhancers['ES_D10'])
 scipy.stats.ranksums(cluster4_enhancers['ES_D7'],cluster4_enhancers['ES_D10'])
 
 
+# Look at Cluster 3 for expression of TF's
+cluster3_tfs = tf_cell_lines.loc[cell_tf_values_std_cluster_3.index.values]
+cluster3_tfs.to_csv("cluster_3_tfs.csv")
+
+box = cluster3_tfs.boxplot(column=['ES_D0', 'ES_D2', 'ES_D5', 'ES_D7',  'ES_D10'],patch_artist=True,showfliers=False,manage_xticks=False,widths = 0.6, medianprops = medianprops)
+plt.setp(box['whiskers'], color='k', linestyle='-', linewidth = 3)
+plt.setp(box['boxes'], color='k', linestyle='-', linewidth = 3)
+
+for patch, color in zip(box['boxes'], colors):
+    patch.set_facecolor(color)
+
+plt.tick_params(axis='y', direction='out')
+plt.tick_params(axis='x', direction='out')
+plt.tick_params(top='off', right='off')
+plt.grid(b=False)
+plt.ylim((-5,55))
+plt.xticks([1,2,3,4,5], ['ES_D0', 'ES_D2', 'ES_D5', 'ES_D7',  'ES_D10'])
+plt.savefig('box_plot_cluster_3_tfs_fpkm.png')
+plt.clf()
+
+# Cluster tfs 1 e-3 (NS)
+scipy.stats.ranksums(cluster3_tfs['ES_D0'],cluster3_tfs['ES_D2'])
+scipy.stats.ranksums(cluster3_tfs['ES_D0'],cluster3_tfs['ES_D5'])
+scipy.stats.ranksums(cluster3_tfs['ES_D0'],cluster3_tfs['ES_D7'])
+scipy.stats.ranksums(cluster3_tfs['ES_D0'],cluster3_tfs['ES_D10'])
+
+
+scipy.stats.ranksums(cluster3_tfs['ES_D2'],cluster3_tfs['ES_D5'])
+scipy.stats.ranksums(cluster3_tfs['ES_D2'],cluster3_tfs['ES_D7'])
+scipy.stats.ranksums(cluster3_tfs['ES_D2'],cluster3_tfs['ES_D10'])
+
+scipy.stats.ranksums(cluster3_tfs['ES_D5'],cluster3_tfs['ES_D7'])
+scipy.stats.ranksums(cluster3_tfs['ES_D5'],cluster3_tfs['ES_D10'])
+
+scipy.stats.ranksums(cluster3_tfs['ES_D7'],cluster3_tfs['ES_D10'])
+
+
+
+
+cluster3_motifs = motif_enhancers.loc[cell_tf_values_std_cluster_3.index.values]
+cluster3_enhancers = only_rpkm_values.loc[cluster3_motifs.loc[:, (cluster3_motifs != 0).all(axis=0)].columns.values]
+cluster3_enhancers.to_csv("cluster_3_enhancers.csv")
+
+box = cluster3_enhancers.boxplot(column=['ES_D0', 'ES_D2', 'ES_D5', 'ES_D7',  'ES_D10'],patch_artist=True,showfliers=False,manage_xticks=False,widths = 0.6, medianprops = medianprops)
+plt.setp(box['whiskers'], color='k', linestyle='-', linewidth = 3)
+plt.setp(box['boxes'], color='k', linestyle='-', linewidth = 3)
+
+for patch, color in zip(box['boxes'], colors):
+    patch.set_facecolor(color)
+
+plt.tick_params(axis='y', direction='out')
+plt.tick_params(axis='x', direction='out')
+plt.tick_params(top='off', right='off')
+plt.grid(b=False)
+plt.ylim((-5,65))
+plt.xticks([1,2,3,4,5], ['ES_D0', 'ES_D2', 'ES_D5', 'ES_D7',  'ES_D10'])
+plt.savefig('box_plot_cluster_3_enhancers_rpkm.png')
+plt.clf()
+
+# Cluster tfs 1 e-12
+scipy.stats.ranksums(cluster3_enhancers['ES_D0'],cluster3_enhancers['ES_D2'])
+scipy.stats.ranksums(cluster3_enhancers['ES_D0'],cluster3_enhancers['ES_D5'])
+scipy.stats.ranksums(cluster3_enhancers['ES_D0'],cluster3_enhancers['ES_D7'])
+scipy.stats.ranksums(cluster3_enhancers['ES_D0'],cluster3_enhancers['ES_D10'])
+
+
+scipy.stats.ranksums(cluster3_enhancers['ES_D2'],cluster3_enhancers['ES_D5'])
+scipy.stats.ranksums(cluster3_enhancers['ES_D2'],cluster3_enhancers['ES_D7'])
+scipy.stats.ranksums(cluster3_enhancers['ES_D2'],cluster3_enhancers['ES_D10'])
+
+scipy.stats.ranksums(cluster3_enhancers['ES_D5'],cluster3_enhancers['ES_D7'])
+scipy.stats.ranksums(cluster3_enhancers['ES_D5'],cluster3_enhancers['ES_D10'])
+
+scipy.stats.ranksums(cluster3_enhancers['ES_D7'],cluster3_enhancers['ES_D10'])
+
 
 ## Analysis of only RNA-seq
 # 1. Z-score Standardize for each cell line to see important TF's

GRO_seq_TFSEE/rank_order.py

+import numpy as np
+import pandas as pd
+import csv
+import matplotlib.pyplot as plt
+
+
+
+
+tfsee = pd.read_csv('clustering_tfs.csv')
+
+tfsee_cluster4 = tfsee[tfsee['cluster'] == 3]
+
+
+tfsee_cluster4['early'] = tfsee_cluster4[['ES_D0','ES_D2','ES_D5']].mean(axis=1)
+tfsee_cluster4['late'] = tfsee_cluster4[['ES_D7','ES_D10']].mean(axis=1)
+tfsee_cluster4['diff'] = tfsee_cluster4['late'] - tfsee_cluster4['early']
+tfsee_cluster4['rank'] = tfsee_cluster4['diff'].rank()
+
+x = list(tfsee_cluster4['rank'])
+
+z = np.polyfit(tfsee_cluster4['rank'], tfsee_cluster4['diff'], 3)
+f = np.poly1d(z)
+x_new = np.linspace(1, 36, num=len(x)*10)
+
+
+plt.figure(figsize=(25,20))
+plt.plot(x_new, f(x_new), color = 'k', linewidth=4.0)
+plt.scatter(x=tfsee_cluster4['rank'], y=tfsee_cluster4['diff'], color='#3999CA', s=600)
+plt.ylim([-0.5,3.5])
+plt.suptitle('Late Differentiation Enriched TFS', fontsize=8, fontweight='bold')
+plt.xlabel('Rank Order',fontsize=8, fontweight='bold')
+plt.ylabel('delta z',fontsize=8, fontweight='bold')
+plt.savefig('cluster4_enriched_tfs.png')
+plt.clf()
+
+
+
+tfsee_cluster3 = tfsee[tfsee['cluster'] == 2]
+
+
+tfsee_cluster3['early'] = tfsee_cluster3[['ES_D0','ES_D2','ES_D5']].mean(axis=1)
+tfsee_cluster3['late'] = tfsee_cluster3[['ES_D7','ES_D10']].mean(axis=1)
+tfsee_cluster3['diff'] = tfsee_cluster3['early'] - tfsee_cluster3['late']
+tfsee_cluster3['rank'] = tfsee_cluster3['diff'].rank()
+
+x = list(tfsee_cluster3['rank'])
+
+z = np.polyfit(tfsee_cluster3['rank'], tfsee_cluster3['diff'], 3)
+f = np.poly1d(z)
+x_new = np.linspace(1, 36, num=len(x)*10)
+
+
+plt.figure(figsize=(25,20))
+plt.plot(x_new, f(x_new), color = 'k', linewidth=4.0)
+plt.scatter(x=tfsee_cluster3['rank'], y=tfsee_cluster3['diff'], color='#316A45', s=600)
+plt.ylim([-0.5,3.5])
+plt.suptitle('Early Enriched TFS', fontsize=8, fontweight='bold')
+plt.xlabel('Rank Order',fontsize=8, fontweight='bold')
+plt.ylabel('delta z',fontsize=8, fontweight='bold')
+plt.savefig('cluster3_enriched_tfs.png')
+plt.clf()