statistics.py
16.7 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
#!/usr/bin/python3.8
import os, pickle
import numpy as np
import pandas as pd
import threading as th
import scipy.stats as st
import matplotlib.pyplot as plt
import matplotlib.cm as cm
import matplotlib.patches as mpatches
import scipy.cluster.hierarchy as sch
from scipy.spatial.distance import squareform
from mpl_toolkits.mplot3d import axes3d
from Bio.Phylo.TreeConstruction import DistanceCalculator
from Bio import AlignIO, SeqIO
from tqdm import tqdm
from functools import partial
from multiprocessing import Pool
from os import path
from collections import Counter
from RNAnet import read_cpu_number
if os.path.isdir("/home/ubuntu/"): # this is the IFB-core cloud
path_to_3D_data = "/mnt/Data/RNA/3D/"
path_to_seq_data = "/mnt/Data/RNA/sequences/"
elif os.path.isdir("/home/persalteas"): # this is my personal workstation
path_to_3D_data = "/home/persalteas/Data/RNA/3D/"
path_to_seq_data = "/home/persalteas/Data/RNA/sequences/"
elif os.path.isdir("/home/lbecquey"): # this is the IBISC server
path_to_3D_data = "/home/lbecquey/Data/RNA/3D/"
path_to_seq_data = "/home/lbecquey/Data/RNA/sequences/"
elif os.path.isdir("/nhome/siniac/lbecquey"): # this is the office PC
path_to_3D_data = "/nhome/siniac/lbecquey/Data/RNA/3D/"
path_to_seq_data = "/nhome/siniac/lbecquey/Data/RNA/sequences/"
else:
print("I don't know that machine... I'm shy, maybe you should introduce yourself ?")
exit(1)
class DataPoint():
def __init__(self, path_to_textfile):
self.df = pd.read_csv(path_to_textfile, sep=',', header=0, engine="c", index_col=0)
self.family = path_to_textfile.split('.')[-1]
self.chain_label = path_to_textfile.split('.')[-2].split('/')[-1]
def load_rna_frome_file(path_to_textfile):
return DataPoint(path_to_textfile)
def reproduce_wadley_results(points, show=False, carbon=4, sd_range=(1,4)):
"""
Plot the joint distribution of pseudotorsion angles, in a Ramachandran-style graph.
See Wadley & Pyle (2007)
Arguments:
show: True or False, call plt.show() at this end or not
filter_helical: None, "form", "zone", or "both"
None: do not remove helical nucleotide
"form": remove nucleotides if they belong to a A, B or Z form stem
"zone": remove nucleotides falling in an arbitrary zone (see zone argument)
"both": remove nucleotides fulfilling one or both of the above conditions
carbon: 1 or 4, use C4' (eta and theta) or C1' (eta_prime and theta_prime)
sd_range: tuple, set values below avg + sd_range[0] * stdev to 0,
and values above avg + sd_range[1] * stdev to avg + sd_range[1] * stdev.
This removes noise and cuts too high peaks, to clearly see the clusters.
"""
worker_nbr = 1 + (carbon==1)
if carbon == 4:
angle = "eta"
xlabel = "$\\eta=C_4'^{i-1}-P^i-C_4'^i-P^{i+1}$"
ylabel = "$\\theta=P^i-C_4'^i-P^{i+1}-C_4'^{i+1}$"
elif carbon == 1:
angle = "eta_prime"
xlabel = "$\\eta'=C_1'^{i-1}-P^i-C_1'^i-P^{i+1}$"
ylabel = "$\\theta'=P^i-C_1'^i-P^{i+1}-C_1'^{i+1}$"
else:
exit("You overestimate my capabilities !")
if not path.isfile(f"data/wadley_kernel_{angle}.npz"):
c2_endo_etas = []
c3_endo_etas = []
c2_endo_thetas = []
c3_endo_thetas = []
for p in points:
df = p.df.loc[(p.df[angle].isna()==False) & (p.df["th"+angle].isna()==False), ["form","puckering", angle,"th"+angle]]
c2_endo_etas += list(df.loc[ (df.puckering=="C2'-endo"), angle ].values)
c3_endo_etas += list(df.loc[ (df.form=='.') & (df.puckering=="C3'-endo"), angle ].values)
c2_endo_thetas += list(df.loc[ (df.puckering=="C2'-endo"), "th"+angle ].values)
c3_endo_thetas += list(df.loc[ (df.form=='.') & (df.puckering=="C3'-endo"), "th"+angle ].values)
xx, yy = np.mgrid[0:2*np.pi:100j, 0:2*np.pi:100j]
positions = np.vstack([xx.ravel(), yy.ravel()])
values_c3 = np.vstack([c3_endo_etas, c3_endo_thetas])
kernel_c3 = st.gaussian_kde(values_c3)
f_c3 = np.reshape(kernel_c3(positions).T, xx.shape)
values_c2 = np.vstack([c2_endo_etas, c2_endo_thetas])
kernel_c2 = st.gaussian_kde(values_c2)
f_c2 = np.reshape(kernel_c2(positions).T, xx.shape)
# Uncomment to save the data to an archive for later use without the need to recompute
np.savez(f"data/wadley_kernel_{angle}.npz",
c3_endo_e=c3_endo_etas, c3_endo_t=c3_endo_thetas,
c2_endo_e=c2_endo_etas, c2_endo_t=c2_endo_thetas,
kernel_c3=f_c3, kernel_c2=f_c2)
else:
f = np.load(f"data/wadley_kernel_{angle}.npz")
c2_endo_etas = f["c2_endo_e"]
c3_endo_etas = f["c3_endo_e"]
c2_endo_thetas = f["c2_endo_t"]
c3_endo_thetas = f["c3_endo_t"]
f_c3 = f["kernel_c3"]
f_c2 = f["kernel_c2"]
xx, yy = np.mgrid[0:2*np.pi:100j, 0:2*np.pi:100j]
print(f"[{worker_nbr}]\tKernel computed (or loaded from file).")
# exact counts:
hist_c2, xedges, yedges = np.histogram2d(c2_endo_etas, c2_endo_thetas, bins=int(2*np.pi/0.1), range=[[0, 2*np.pi], [0, 2*np.pi]])
hist_c3, xedges, yedges = np.histogram2d(c3_endo_etas, c3_endo_thetas, bins=int(2*np.pi/0.1), range=[[0, 2*np.pi], [0, 2*np.pi]])
color_values = cm.jet(hist_c3.ravel()/hist_c3.max())
for x, y, hist, f, l in zip( (c3_endo_etas, c2_endo_etas), (c3_endo_thetas, c2_endo_thetas), (hist_c3, hist_c2), (f_c3, f_c2), ("c3","c2")):
# cut hist and kernel
hist_sup_thr = hist.mean() + sd_range[1]*hist.std()
hist_cut = np.where( hist > hist_sup_thr, hist_sup_thr, hist)
f_sup_thr = f.mean() + sd_range[1]*f.std()
f_low_thr = f.mean() + sd_range[0]*f.std()
f_cut = np.where(f > f_sup_thr, f_sup_thr, f)
f_cut = np.where(f_cut < f_low_thr, 0, f_cut)
levels = [f.mean()+f.std(), f.mean()+2*f.std(), f.mean()+4*f.std()]
# histogram:
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
xpos, ypos = np.meshgrid(xedges[:-1], yedges[:-1], indexing="ij")
ax.bar3d(xpos.ravel(), ypos.ravel(), 0.0, 0.09, 0.09, hist_cut.ravel(), color=color_values, zorder="max")
ax.set_xlabel(xlabel)
ax.set_ylabel(ylabel)
plt.savefig(f"results/wadley_hist_{angle}_{l}.png")
if show:
plt.show()
plt.close()
# Smoothed joint distribution
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.plot_surface(xx, yy, f_cut, cmap=cm.coolwarm, linewidth=0, antialiased=True)
ax.set_xlabel(xlabel)
ax.set_ylabel(ylabel)
plt.savefig(f"results/wadley_distrib_{angle}_{l}.png")
if show:
plt.show()
plt.close()
# 2D Wadley plot
fig = plt.figure(figsize=(5,5))
ax = fig.gca()
ax.scatter(x, y, s=1, alpha=0.1)
ax.contourf(xx, yy, f_cut, alpha=0.5, cmap=cm.coolwarm, levels=levels, extend="max")
ax.set_xlabel(xlabel)
ax.set_ylabel(ylabel)
fig.savefig(f"results/wadley_{angle}_{l}.png")
if show:
plt.show()
print(f"[{worker_nbr}]\tComputed joint distribution of angles (C{carbon}) and saved the figures.")
def stats_len(mappings_list, points):
cols = []
lengths = []
for f in sorted(mappings_list.keys()):
if f in ["RF02540","RF02541","RF02543"]:
cols.append("red") # LSU
elif f in ["RF00177","RF01960","RF01959","RF02542"]:
cols.append("blue") # SSU
elif f in ["RF00001"]:
cols.append("green")
elif f in ["RF00002"]:
cols.append("purple")
elif f in ["RF00005"]:
cols.append("orange")
else:
cols.append("grey")
l = []
for r in points:
if r.family != f: continue
l.append(len(r.df['nt_code']))
lengths.append(l)
plt.figure(figsize=(10,3))
ax = plt.gca()
ax.hist(lengths, bins=100, stacked=True, log=True, color=cols, label=sorted(mappings_list.keys()))
ax.set_xlabel("Sequence length (nucleotides)")
ax.set_ylabel("Number of 3D chains")
plt.tight_layout()
handles, labels = ax.get_legend_handles_labels()
filtered_handles = [mpatches.Patch(color='red'), mpatches.Patch(color='white'),
mpatches.Patch(color='blue'), mpatches.Patch(color='white'),
mpatches.Patch(color='green'), mpatches.Patch(color='purple'),
mpatches.Patch(color='orange'), mpatches.Patch(color='grey')]
filtered_labels = ['Large Ribosomal Subunits', '(RF02540, RF02541, RF02543)','Small Ribosomal Subunits','(RF01960, RF00177)',
'5S rRNA (RF00001)', '5.8S rRNA (RF00002)', 'tRNA (RF00005)', 'Other']
ax.legend(filtered_handles, filtered_labels, loc='best', ncol=2)# bbox_to_anchor=(0.5, -0.5), ncol=4, fontsize=)
plt.savefig("results/lengths.png")
print("[3]\tComputed sequence length statistics and saved the figure.")
def format_percentage(tot, x):
if not tot:
return '0 %'
x = 100*x/tot
if x >= 0.01:
x = "%.2f" % x
else:
x = "<.01"
return x + '%'
def stats_freq(mappings_list, points):
freqs = {}
for f in mappings_list.keys():
freqs[f] = Counter()
for r in points:
freqs[r.family].update(dict(r.df['nt_name'].value_counts()))
df = pd.DataFrame()
for f in sorted(mappings_list.keys()):
tot = sum(freqs[f].values())
df = pd.concat([ df, pd.DataFrame([[ format_percentage(tot, x) for x in freqs[f].values() ]], columns=list(freqs[f]), index=[f]) ])
df = df.fillna(0)
df.to_csv("results/frequencies.csv")
print("[4]\tComputed nucleotide statistics and saved CSV file.")
def stats_pairs(mappings_list, points):
def line_format(family_data):
return family_data.apply(partial(format_percentage, sum(family_data)))
# Create a Counter() object by family
freqs = {}
for f in mappings_list.keys():
freqs[f] = Counter()
# Iterate over data points
for r in tqdm(points, desc="RNA points", position=0, leave=False):
# Skip if linear piece of RNA
if not sum([ x != 0 for x in r.df.paired ]):
continue
# Count each pair type within the molecule
vcnts = pd.concat(
[ pd.Series(row['pair_type_LW'].split(','))
for _, row in r.df.dropna(subset=["pair_type_LW"]).iterrows() ]
).reset_index(drop=True).value_counts()
# Add these new counts to the family's counter
freqs[r.family].update(dict(vcnts))
# Create the output dataframe
df = pd.DataFrame()
for f in sorted(mappings_list.keys()):
df = pd.concat([ df, pd.DataFrame([[ x for x in freqs[f].values() ]], columns=list(freqs[f]), index=[f]) ])
df = df.fillna(0)
# Remove not very well defined pair types (not in the 12 LW types)
col_list = [ x for x in df.columns if '.' in x ]
df['other'] = df[col_list].sum(axis=1)
df.drop(col_list, axis=1, inplace=True)
# Compute total row
total_series = df.sum(numeric_only=True).rename("TOTAL")
df = df.append(total_series)
# format as percentages
df = df.apply(line_format, axis=1)
# reorder columns
df.sort_values("TOTAL", axis=1, inplace=True, ascending=False)
# Save to CSV
df.to_csv("results/pairings.csv")
# Plot barplot of overall types
total_series.sort_values(ascending=False, inplace=True)
total_series.apply(lambda x: x/2.0) # each interaction was counted twice because one time per extremity
ax = total_series.plot(figsize=(5,3), kind='bar', log=True, ylim=(1e4,5000000) )
ax.set_ylabel("Number of observations")
plt.subplots_adjust(bottom=0.2, right=0.99)
plt.savefig("results/pairings.png")
print("[5]\tComputed nucleotide statistics and saved CSV and PNG file.")
def to_dist_matrix(f):
if path.isfile("data/"+f+".npy"):
return 0
dm = DistanceCalculator('identity')
with open(path_to_seq_data+"realigned/"+f+"++.afa") as al_file:
al = AlignIO.read(al_file, "fasta")[-len(mappings_list[f]):]
idty = dm.get_distance(al).matrix # list of lists
del al
l = len(idty)
np.save("data/"+f+".npy", np.array([ idty[i] + [0]*(l-1-i) if i<l-1 else idty[i] for i in range(l) ]))
del idty
return 0
def seq_idty(mappings_list):
famlist = sorted([ x for x in mappings_list.keys() if len(mappings_list[x]) > 1 ])
ignored = []
for x in mappings_list.keys():
if len(mappings_list[x]) == 1:
ignored.append(x)
if len(ignored):
print("Ignoring families with only one chain:", " ".join(ignored))
# compute distance matrices
p = Pool(processes=8)
pbar = tqdm(total=len(famlist), desc="Families idty matrices", position=1, leave=True)
for i, _ in enumerate(p.imap_unordered(to_dist_matrix, famlist)):
pbar.update(1)
pbar.close()
p.close()
p.join()
# load them
fam_arrays = []
for f in famlist:
if path.isfile("data/"+f+".npy"):
fam_arrays.append(np.load("data/"+f+".npy"))
else:
fam_arrays.append([])
fig, axs = plt.subplots(5,13, figsize=(15,9))
axs = axs.ravel()
[axi.set_axis_off() for axi in axs]
for f, D, ax in zip(famlist, fam_arrays, axs):
if not len(D): continue
if D.shape[0] > 2: # Cluster only if there is more than 2 sequences to organize
D = D + D.T # Copy the lower triangle to upper, to get a symetrical matrix
condensedD = squareform(D)
# Compute basic dendrogram by Ward's method
Y = sch.linkage(condensedD, method='ward')
Z = sch.dendrogram(Y, orientation='left', no_plot=True)
# Reorganize rows and cols
idx1 = Z['leaves']
D = D[idx1,:]
D = D[:,idx1[::-1]]
im = ax.matshow(1.0 - D, vmin=0, vmax=1, origin='lower') # convert to identity matrix 1 - D from distance matrix D
ax.set_title(f + "\n(" + str(len(mappings_list[f]))+ " chains)")
fig.tight_layout()
fig.subplots_adjust(wspace=0.1, hspace=0.3)
fig.colorbar(im, ax=axs[-1], shrink=0.8)
fig.savefig(f"results/distances.png")
print("[6]\tComputed identity matrices and saved the figure.")
if __name__ == "__main__":
#################################################################
# LOAD ALL FILES
#################################################################
print("Loading mappings list...")
mappings_list = pd.read_csv(path_to_seq_data + "realigned/mappings_list.csv", sep=',', index_col=0).to_dict(orient='list')
for k in mappings_list.keys():
mappings_list[k] = [ x for x in mappings_list[k] if str(x) != 'nan' ]
print("Loading datapoints from file...")
if path.isfile("data/rnapoints.dat"):
with open("data/rnapoints.dat", 'rb') as f:
rna_points = pickle.load(f)
else:
rna_points = []
filelist = [path_to_3D_data+"/datapoints/"+f for f in os.listdir(path_to_3D_data+"/datapoints") if ".log" not in f and ".gz" not in f]
p = Pool(initializer=tqdm.set_lock, initargs=(tqdm.get_lock(),), processes=read_cpu_number())
pbar = tqdm(total=len(filelist), desc="RNA files", position=0, leave=True)
for i, rna in enumerate(p.imap_unordered(load_rna_frome_file, filelist)):
rna_points.append(rna)
pbar.update(1)
pbar.close()
p.close()
p.join()
with open("data/rnapoints.dat", "wb") as f:
pickle.dump(rna_points, f)
npoints = len(rna_points)
print(npoints, "RNA files loaded.")
#################################################################
# Define threads for the tasks
#################################################################
wadley_thr = []
wadley_thr.append(th.Thread(target=reproduce_wadley_results, args=[rna_points], kwargs={'carbon': 1}))
wadley_thr.append(th.Thread(target=reproduce_wadley_results, args=[rna_points], kwargs={'carbon': 4}))
seq_len_thr = th.Thread(target=partial(stats_len, mappings_list), args=[rna_points])
nt_freq_thr = th.Thread(target=partial(stats_freq, mappings_list), args=[rna_points])
pairs_freq_thr = th.Thread(target=partial(stats_pairs, mappings_list), args=[rna_points])
dist_thr = th.Thread(target=seq_idty, args=[mappings_list])
for t in wadley_thr:
t.start()
seq_len_thr.start()
nt_freq_thr.start()
pairs_freq_thr.start()
dist_thr.start()
for t in wadley_thr:
t.join()
seq_len_thr.join()
nt_freq_thr.join()
pairs_freq_thr.join()
dist_thr.join()