def compare_entropy(G):
FILE = "/Users/joazofeifa/Lab/TF_predictions/HOCOMOCOv9_AD_MEME.txt"
D = at.load_PSSMs(FILE,test=False)
xy = list()
for exp in G:
if np.random.uniform(0,1) < 0.5:
for TF in G[exp]:
GC = get_GC_motif(D, TF)
if GC and G[exp][TF][0]:
xy.append((G[exp][TF][0],G[exp][TF][3], GC) )
F = plt.figure()
ax = F.add_axes([0.1, 0.1, 0.35, 0.35])
axc = F.add_axes([0.5, 0.1, 0.05, 0.35])
ax3 = F.add_axes([0.6, 0.55, 0.35, 0.35])
ax2 = F.add_axes([0.1, 0.55, 0.35, 0.35])
cmap = plt.get_cmap('Blues')
x,y = [math.log(u, 10) for u,v,z in xy if v > 0.005 ],[v for u,v,z in xy if v> 0.005 ]
c = [z for u,v,z in xy if v > 0.005 ]
print len(c), len(y)
norm = mpl.colors.Normalize(vmin=min(c), vmax=max(c))
m = cm.ScalarMappable(norm=norm, cmap=cmap)
cc = [m.to_rgba(C) for C in c]
cb1 = mpl.colorbar.ColorbarBase(axc, cmap=cmap,
norm=norm,
orientation='vertical')
cb1.set_label('GC Content')
ax.scatter(x,y, color=cc, edgecolor='' )
ax.set_xlabel("KS-Test")
ax.set_ylabel("True Positive Rate")
ax.grid()
ax2.hist(c, bins=25)
ax2.set_xlabel("GC Content for All HOCOMOCO Motifs")
ax2.set_ylabel("Frequency")
XY = np.vstack([c,y])
z = gaussian_kde(XY)(XY)
ax3.scatter(c,y, s=5, edgecolor='')
ax3.set_xlabel("GC Content")
ax3.set_ylabel("True Positive Rate")
ax3.grid()
plt.show()
pass
nx.draw_networkx_edges(G, pos=pos, alpha=0.7,width=3,ax=ax,edge_color=colors) nx.draw_networkx_labels(G,pos,LABELS,font_size=7, ax=ax) plt.show() if __name__ == "__main__": make_distance = False OUT = "/Users/joazofeifa/Lab/EMG/TF_predictions/files/" SCIPY = False CUSTOM = False ts, xs = get_counts(OUT+"thresholds_nearest_assignments.tsv", show_thresh=False) threshold_as_network(xs[5]) if make_distance: FILE = "/Users/joazofeifa/Lab/TF_predictions/HOCOMOCOv9_AD_MEME.txt" D = at.load_PSSMs(FILE,test=False) perform_linkage(D, OUT=OUT) if CUSTOM: FILE = "/Users/joazofeifa/Lab/TF_predictions/HOCOMOCOv9_AD_MEME.txt" KL,M = load_matrix(OUT+"kl_distance_matrix.csv") try_different_thresholds(KL, M, OUT=OUT, res=50) if SCIPY: KL,M = load_matrix(OUT+"kl_distance_matrix.csv") H,M = load_matrix(OUT+"hamming_distance_matrix.csv") SCIPY = True if SCIPY: P = linkage(KL,H,M, SHOW=True) extract_clusters(P, threshold=None)