Mercurial > repos > mheinzl > fsd

comparison fsd.py @ 41:54f0dac1c834 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/monikaheinzl/duplexanalysis_galaxy/tree/master/tools/fsd commit 033dd7b750f68e8aa68f327d7d72bd311ddbee4e-dirty
author mheinzl
date Wed, 14 Aug 2019 04:30:58 -0400
parents 00f5ca6bb610
children 321a4871564b
comparison
equal deleted inserted replaced
40:948064d8714e 41:54f0dac1c834
112 112
113 legend3b = "PE reads\n{:,} ({:.3f})".format(numpy.bincount(data1)[1], float(numpy.bincount(data1)[1]) / sum(integers)) 113 legend3b = "PE reads\n{:,} ({:.3f})".format(numpy.bincount(data1)[1], float(numpy.bincount(data1)[1]) / sum(integers))
114 fig.text(0.45, 0.11, legend3b, size=10, transform=plt.gcf().transFigure) 114 fig.text(0.45, 0.11, legend3b, size=10, transform=plt.gcf().transFigure)
115 fig2.text(0.45, 0.11, legend3b, size=10, transform=plt.gcf().transFigure) 115 fig2.text(0.45, 0.11, legend3b, size=10, transform=plt.gcf().transFigure)
116 116
117 legend4 = "family size > 20:\nnr. of tags\n{:,} ({:.3f})".format(numpy.bincount(data1)[len(numpy.bincount(data1)) - 1].astype(int), float(numpy.bincount(data1)[len(numpy.bincount(data1)) - 1]) / len(data1)) 117 legend4 = "family size > 20:\nnr. of tags\n{:,} ({:.3f})".format(len(integers[integers > 20]),
118 float(sum(integers[integers > 20]))
119 / sum(integers))
118 fig.text(0.58, 0.11, legend4, size=10, transform=plt.gcf().transFigure) 120 fig.text(0.58, 0.11, legend4, size=10, transform=plt.gcf().transFigure)
119 fig2.text(0.58, 0.11, legend4, size=10, transform=plt.gcf().transFigure) 121 fig2.text(0.58, 0.11, legend4, size=10, transform=plt.gcf().transFigure)
120 122
121 legend5 = "PE reads\n{:,} ({:.3f})".format(sum(integers[integers > 20]), float(sum(integers[integers > 20])) / sum(integers)) 123 legend5 = "PE reads\n{:,} ({:.3f})".format(sum(integers[integers > 20]), float(sum(integers[integers > 20])) / sum(integers))
122 fig.text(0.70, 0.11, legend5, size=10, transform=plt.gcf().transFigure) 124 fig.text(0.70, 0.11, legend5, size=10, transform=plt.gcf().transFigure)
155 157
156 legend3 = "{:,} ({:.3f})".format(numpy.bincount(data2)[1], float(numpy.bincount(data2)[1]) / sum(integers2)) 158 legend3 = "{:,} ({:.3f})".format(numpy.bincount(data2)[1], float(numpy.bincount(data2)[1]) / sum(integers2))
157 fig.text(0.45, 0.09, legend3, size=10, transform=plt.gcf().transFigure) 159 fig.text(0.45, 0.09, legend3, size=10, transform=plt.gcf().transFigure)
158 fig2.text(0.45, 0.09, legend3, size=10, transform=plt.gcf().transFigure) 160 fig2.text(0.45, 0.09, legend3, size=10, transform=plt.gcf().transFigure)
159 161
160 legend4 = "{:,} ({:.3f})".format( 162 legend4 = "{:,} ({:.3f})".format(len(integers2[integers2 > 20]), float(sum(integers2[integers2 > 20])) / sum(integers2))
161 numpy.bincount(data2)[len(numpy.bincount(data2)) - 1].astype(int),
162 float(numpy.bincount(data2)[len(numpy.bincount(data2)) - 1]) / len(data2))
163 fig.text(0.58, 0.09, legend4, size=10, transform=plt.gcf().transFigure) 163 fig.text(0.58, 0.09, legend4, size=10, transform=plt.gcf().transFigure)
164 fig2.text(0.58, 0.09, legend4, size=10, transform=plt.gcf().transFigure) 164 fig2.text(0.58, 0.09, legend4, size=10, transform=plt.gcf().transFigure)
165 165
166 legend5 = "{:,} ({:.3f})".format(sum(integers2[integers2 > 20]), float(sum(integers2[integers2 > 20])) / sum(integers2)) 166 legend5 = "{:,} ({:.3f})".format(sum(integers2[integers2 > 20]), float(sum(integers2[integers2 > 20])) / sum(integers2))
167 fig.text(0.70, 0.09, legend5, size=10, transform=plt.gcf().transFigure) 167 fig.text(0.70, 0.09, legend5, size=10, transform=plt.gcf().transFigure)
200 200
201 legend3b = "{:,} ({:.3f})".format(numpy.bincount(data3)[1], float(numpy.bincount(data3)[1]) / sum(integers3)) 201 legend3b = "{:,} ({:.3f})".format(numpy.bincount(data3)[1], float(numpy.bincount(data3)[1]) / sum(integers3))
202 fig.text(0.45, 0.07, legend3b, size=10, transform=plt.gcf().transFigure) 202 fig.text(0.45, 0.07, legend3b, size=10, transform=plt.gcf().transFigure)
203 fig2.text(0.45, 0.07, legend3b, size=10, transform=plt.gcf().transFigure) 203 fig2.text(0.45, 0.07, legend3b, size=10, transform=plt.gcf().transFigure)
204 204
205 legend4 = "{:,} ({:.3f})".format( 205 legend4 = "{:,} ({:.3f})".format(len(integers3[integers3 > 20]), float(sum(integers3[integers3 > 20])) / sum(integers3))
206 numpy.bincount(data3)[len(numpy.bincount(data3)) - 1].astype(int),
207 float(numpy.bincount(data3)[len(numpy.bincount(data3)) - 1]) / len(data3))
208 fig.text(0.58, 0.07, legend4, size=10, transform=plt.gcf().transFigure) 206 fig.text(0.58, 0.07, legend4, size=10, transform=plt.gcf().transFigure)
209 fig2.text(0.58, 0.07, legend4, size=10, transform=plt.gcf().transFigure) 207 fig2.text(0.58, 0.07, legend4, size=10, transform=plt.gcf().transFigure)
210 208
211 legend5 = "{:,} ({:.3f})".format(sum(integers3[integers3 > 20]), 209 legend5 = "{:,} ({:.3f})".format(sum(integers3[integers3 > 20]),
212 float(sum(integers3[integers3 > 20])) / sum(integers3)) 210 float(sum(integers3[integers3 > 20])) / sum(integers3))
245 243
246 legend3b = "{:,} ({:.3f})".format(numpy.bincount(data4)[1], float(numpy.bincount(data4)[1]) / sum(integers4)) 244 legend3b = "{:,} ({:.3f})".format(numpy.bincount(data4)[1], float(numpy.bincount(data4)[1]) / sum(integers4))
247 fig.text(0.45, 0.05, legend3b, size=10, transform=plt.gcf().transFigure) 245 fig.text(0.45, 0.05, legend3b, size=10, transform=plt.gcf().transFigure)
248 fig2.text(0.45, 0.05, legend3b, size=10, transform=plt.gcf().transFigure) 246 fig2.text(0.45, 0.05, legend3b, size=10, transform=plt.gcf().transFigure)
249 247
250 legend4 = "{:,} ({:.3f})".format( 248 legend4 = "{:,} ({:.3f})".format(len(integers4[integers4 > 20]), float(sum(integers4[integers4 > 20])) / sum(integers4))
251 numpy.bincount(data4)[len(numpy.bincount(data4)) - 1].astype(int),
252 float(numpy.bincount(data4)[len(numpy.bincount(data4)) - 1]) / len(data4))
253 fig.text(0.58, 0.05, legend4, size=10, transform=plt.gcf().transFigure) 249 fig.text(0.58, 0.05, legend4, size=10, transform=plt.gcf().transFigure)
254 fig2.text(0.58, 0.05, legend4, size=10, transform=plt.gcf().transFigure) 250 fig2.text(0.58, 0.05, legend4, size=10, transform=plt.gcf().transFigure)
255 251
256 legend5 = "{:,} ({:.3f})".format(sum(integers4[integers4 > 20]), 252 legend5 = "{:,} ({:.3f})".format(sum(integers4[integers4 > 20]),
257 float(sum(integers4[integers4 > 20])) / sum(integers4)) 253 float(sum(integers4[integers4 > 20])) / sum(integers4))
284 if to_plot[l] == "Relative frequencies": 280 if to_plot[l] == "Relative frequencies":
285 w = [numpy.zeros_like(data) + 1. / len(data) for data in list_to_plot2] 281 w = [numpy.zeros_like(data) + 1. / len(data) for data in list_to_plot2]
286 counts_rel = ax.hist(list_to_plot2, weights=w, 282 counts_rel = ax.hist(list_to_plot2, weights=w,
287 bins=numpy.arange(1, 23), stacked=False, edgecolor="black", 283 bins=numpy.arange(1, 23), stacked=False, edgecolor="black",
288 linewidth=1, label=label, align="left", alpha=0.7, rwidth=0.8) 284 linewidth=1, label=label, align="left", alpha=0.7, rwidth=0.8)
285 ax.set_ylim(0, 1.07)
289 else: 286 else:
290 counts = ax.hist(list_to_plot2, bins=numpy.arange(minimumX, maximumX + 2), stacked=False, edgecolor="black", linewidth=1, label=label, align="left", alpha=0.7, rwidth=0.8, color=colors) 287 counts = ax.hist(list_to_plot2, bins=numpy.arange(minimumX, maximumX + 2), stacked=False, edgecolor="black", linewidth=1, label=label, align="left", alpha=0.7, rwidth=0.8, color=colors)
291 ax.legend(loc='upper right', fontsize=14, frameon=True, bbox_to_anchor=(0.9, 1)) 288 ax.legend(loc='upper right', fontsize=14, frameon=True, bbox_to_anchor=(0.9, 1))
292 289
293 ax.set_xticks(numpy.array(ticks)) 290 ax.set_xticks(numpy.array(ticks))
325 else: 322 else:
326 c_idx = numpy.where(t == unique)[0] 323 c_idx = numpy.where(t == unique)[0]
327 new_c.append(c[c_idx]) 324 new_c.append(c[c_idx])
328 new_unique.append(unique[c_idx]) 325 new_unique.append(unique[c_idx])
329 y = numpy.array(new_unique) * numpy.array(new_c) 326 y = numpy.array(new_unique) * numpy.array(new_c)
330 if len([list_to_plot_original > 20]) > 0: 327 if sum(list_to_plot_original[i] > 20) > 0:
331 y[len(y) - 1] = sum(list_to_plot_original[i][list_to_plot_original[i] > 20]) 328 y[len(y) - 1] = sum(list_to_plot_original[i][list_to_plot_original[i] > 20])
329 # y = [y[x[idx] == unique][0] if x[idx] in unique else 0 for idx in range(len(x))]
332 reads.append(y) 330 reads.append(y)
333 reads_rel.append(list(numpy.float_(y)) / sum(y)) 331 reads_rel.append(list(numpy.float_(y)) / sum(y))
334 332
335 x = list(numpy.arange(numpy.amin(unique), numpy.amax(unique) + 1).astype(float)) 333 x = list(numpy.arange(numpy.amin(unique), numpy.amax(unique) + 1).astype(float))
336 if len(list_to_plot2) == 1: 334 if len(list_to_plot2) == 1:
341 w = 1./(len(list_to_plot) + 1) 339 w = 1./(len(list_to_plot) + 1)
342 340
343 if to_plot[l] == "Relative frequencies": 341 if to_plot[l] == "Relative frequencies":
344 counts2_rel = ax.bar(x, list(numpy.float_(y)) / numpy.sum(y), align="edge", width=w, 342 counts2_rel = ax.bar(x, list(numpy.float_(y)) / numpy.sum(y), align="edge", width=w,
345 edgecolor="black", label=label[i],linewidth=1, alpha=0.7, color=colors[i]) 343 edgecolor="black", label=label[i],linewidth=1, alpha=0.7, color=colors[i])
346 344 ax.set_ylim(0, 1.07)
347 else: 345 else:
348 y = list(y.reshape((len(y)))) 346 y = list(y.reshape((len(y))))
349 347
350 counts2 = ax.bar(x, y, align="edge", width=w, edgecolor="black", label=label[i], linewidth=1, 348 counts2 = ax.bar(x, y, align="edge", width=w, edgecolor="black", label=label[i], linewidth=1,
351 alpha=0.7, color=colors[i]) 349 alpha=0.7, color=colors[i])