Mercurial > repos > iuc > b2btools_single_sequence

annotate script.py @ 0:dc91f5626375 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
author iuc
date Tue, 09 Aug 2022 12:29:40 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
1 import json
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
2 import optparse
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
3 import os.path
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
4 import re
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
5 import unicodedata
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
6
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
7 import matplotlib.pyplot as plt
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
8 import numpy as np
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
9 import pandas as pd
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
10 from b2bTools import SingleSeq
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
11
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
12
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
13 def slugify(value):
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
14 """
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
15 From https://github.com/django/django/blob/master/django/utils/text.py
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
16 Convert to ASCII if 'allow_unicode'. Convert spaces or repeated
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
17 dashes to single dashes. Remove characters that aren't alphanumerics,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
18 underscores, or hyphens. Convert to lowercase. Also strip leading and
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
19 trailing whitespace, dashes, and underscores.
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
20 """
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
21 value = str(value)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
22 value = (
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
23 unicodedata.normalize("NFKD", value)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
24 .encode("ascii", "ignore")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
25 .decode("ascii")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
26 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
27 value = re.sub(r"[^\w\s-]", "", value.lower())
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
28 return re.sub(r"[-\s]+", "-", value).strip("-_")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
29
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
30
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
31 def check_min_max(predicted_values, former_min, former_max):
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
32 seq_max = max(predicted_values)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
33 seq_min = min(predicted_values)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
34 if (
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
35 seq_max + 0.1 > former_max
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
36 and not np.isnan(seq_max)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
37 and not np.isinf(seq_max)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
38 ):
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
39 former_max = seq_max + 0.1
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
40 if (
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
41 seq_min - 0.1 < former_min
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
42 and not np.isnan(seq_min)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
43 and not np.isinf(seq_min)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
44 ):
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
45 former_min = seq_min - 0.1
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
46 return former_min, former_max
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
47
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
48
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
49 def plot_prediction(pred_name, hlighting_regions, predicted_values, seq_name):
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
50 thresholds_dict = {
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
51 "backbone": {
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
52 "membrane spanning": [1.0, 1.5],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
53 "rigid": [0.8, 1.0],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
54 "context-dependent": [0.69, 0.8],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
55 "flexible": [-1.0, 0.69],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
56 },
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
57 "earlyFolding": {
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
58 "early folds": [0.169, 2.0],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
59 "late folds": [-1.0, 0.169],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
60 },
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
61 "disoMine": {"ordered": [-1.0, 0.5], "disordered": [0.5, 2.0]},
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
62 }
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
63 ordered_regions_dict = {
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
64 "backbone": [
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
65 "flexible",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
66 "context-dependent",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
67 "rigid",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
68 "membrane spanning",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
69 ],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
70 "earlyFolding": ["late folds", "early folds"],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
71 "disoMine": ["ordered", "disordered"],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
72 }
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
73 colors = ["yellow", "orange", "pink", "red"]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
74 ranges_dict = {
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
75 "backbone": [-0.2, 1.2],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
76 "sidechain": [-0.2, 1.2],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
77 "ppII": [-0.2, 1.2],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
78 "earlyFolding": [-0.2, 1.2],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
79 "disoMine": [-0.2, 1.2],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
80 "agmata": [-0.2, 1.2],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
81 "helix": [-1.0, 1.0],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
82 "sheet": [-1.0, 1.0],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
83 "coil": [-1.0, 1.0],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
84 }
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
85 fig, ax = plt.subplots(1, 1)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
86 fig.set_figwidth(10)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
87 fig.set_figheight(5)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
88 ax.set_title(pred_name + " " + "prediction")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
89 min_value, max_value = ranges_dict[pred_name]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
90 if seq_name == "all":
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
91 max_len = 0
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
92 for seq in predicted_values.keys():
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
93 predictions = predicted_values[seq]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
94 min_value, max_value = check_min_max(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
95 predictions, min_value, max_value
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
96 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
97 ax.plot(range(len(predictions)), predictions, label=seq)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
98 if len(predictions) > max_len:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
99 max_len = len(predictions)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
100 ax.set_xlim([0, max_len - 1])
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
101 else:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
102 predictions = predicted_values
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
103 min_value, max_value = check_min_max(predictions, min_value, max_value)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
104 ax.plot(range(len(predictions)), predictions, label=seq_name)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
105 ax.set_xlim([0, len(predictions) - 1])
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
106 legend_lines = plt.legend(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
107 bbox_to_anchor=(1.04, 1), loc="upper left", fancybox=True, shadow=True
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
108 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
109 ax.add_artist(legend_lines)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
110 # Define regions
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
111 if hlighting_regions:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
112 if pred_name in ordered_regions_dict.keys():
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
113 for i, prediction in enumerate(ordered_regions_dict[pred_name]):
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
114 lower = thresholds_dict[pred_name][prediction][0]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
115 upper = thresholds_dict[pred_name][prediction][1]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
116 color = colors[i]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
117 ax.axhspan(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
118 lower, upper, alpha=0.3, color=color, label=prediction
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
119 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
120 included_in_regions_legend = list(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
121 reversed(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
122 [
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
123 prediction
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
124 for prediction in ordered_regions_dict[pred_name]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
125 ]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
126 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
127 ) # to sort it "from up to low"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
128 # Get handles and labels
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
129 handles, labels = plt.gca().get_legend_handles_labels()
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
130 handles_dict = {
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
131 label: handles[idx] for idx, label in enumerate(labels)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
132 }
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
133 # Add legend for regions, if available
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
134 region_legend = ax.legend(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
135 [
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
136 handles_dict[region]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
137 for region in included_in_regions_legend
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
138 ],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
139 [region for region in included_in_regions_legend],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
140 fancybox=True,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
141 shadow=True,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
142 loc="lower left",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
143 bbox_to_anchor=(1.04, 0),
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
144 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
145 ax.add_artist(region_legend)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
146 ax.set_ylim([min_value, max_value])
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
147 ax.set_xlabel("residue index")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
148 ax.set_ylabel("prediction values")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
149 ax.grid(axis="y")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
150 plt.savefig(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
151 os.path.join(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
152 options.plot_output,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
153 "{0}_{1}.png".format(slugify(seq_name), pred_name),
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
154 ),
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
155 bbox_inches="tight",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
156 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
157 plt.close()
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
158
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
159
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
160 def df_dict_to_dict_of_values(df_dict, predictor):
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
161 results_dict = {}
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
162 for seq in df_dict.keys():
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
163 df = pd.read_csv(df_dict[seq], sep="\t")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
164 results_dict[seq] = df[predictor]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
165 return results_dict
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
166
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
167
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
168 def main(options):
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
169 single_seq = SingleSeq(options.input_fasta)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
170 b2b_tools = []
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
171 if options.dynamine:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
172 b2b_tools.append("dynamine")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
173 if options.disomine:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
174 b2b_tools.append("disomine")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
175 if options.efoldmine:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
176 b2b_tools.append("efoldmine")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
177 if options.agmata:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
178 b2b_tools.append("agmata")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
179 single_seq.predict(b2b_tools)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
180 predictions = single_seq.get_all_predictions()
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
181
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
182 def rounder_function(value):
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
183 return round(float(value), 3)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
184
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
185 rounded_predictions = json.loads(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
186 json.dumps(predictions), parse_float=rounder_function
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
187 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
188 results_json = json.dumps(rounded_predictions, indent=2, sort_keys=True)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
189 with open(options.json_output, "w") as f:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
190 f.write(results_json)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
191 first_sequence_key = next(iter(predictions))
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
192 prediction_keys = predictions[first_sequence_key].keys()
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
193 # Sort column names
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
194 tsv_column_names = list(prediction_keys)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
195 tsv_column_names.remove("seq")
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
196 tsv_column_names = ['residue', *sorted(tsv_column_names)]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
197
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
198 df_dictionary = {}
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
199 for sequence_key, seq_preds in predictions.items():
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
200 residues = seq_preds["seq"]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
201 residues_count = len(residues)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
202 sequence_df = pd.DataFrame(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
203 columns=prediction_keys, index=range(residues_count)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
204 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
205 sequence_df.index.name = "residue_index"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
206 for predictor in prediction_keys:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
207 sequence_df[predictor] = seq_preds[predictor]
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
208 sequence_df = sequence_df.rename(columns={"seq": "residue"})
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
209 sequence_df = sequence_df.round(decimals=3)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
210 filename = f"{options.output}/{slugify(sequence_key)}.tsv"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
211 df_dictionary[sequence_key] = filename
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
212 sequence_df.to_csv(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
213 filename,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
214 header=True,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
215 columns=tsv_column_names,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
216 sep="\t"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
217 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
218 # Plot each individual plot (compatible with plot all)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
219 if options.plot:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
220 for predictor in prediction_keys:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
221 if predictor != "seq":
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
222 plot_prediction(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
223 pred_name=predictor,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
224 hlighting_regions=True,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
225 predicted_values=seq_preds[predictor],
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
226 seq_name=sequence_key,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
227 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
228 # Plot all together (compatible with plot individual)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
229 if options.plot_all:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
230 for predictor in prediction_keys:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
231 if predictor != "seq":
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
232 results_dictionary = df_dict_to_dict_of_values(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
233 df_dictionary, predictor
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
234 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
235 plot_prediction(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
236 pred_name=predictor,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
237 hlighting_regions=True,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
238 predicted_values=results_dictionary,
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
239 seq_name="all",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
240 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
241
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
242
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
243 if __name__ == "__main__":
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
244 parser = optparse.OptionParser()
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
245 parser.add_option(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
246 "--dynamine",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
247 action="store_true"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
248 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
249 parser.add_option(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
250 "--disomine",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
251 action="store_true"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
252 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
253 parser.add_option(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
254 "--efoldmine",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
255 action="store_true"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
256 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
257 parser.add_option(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
258 "--agmata",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
259 action="store_true"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
260 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
261 parser.add_option(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
262 "--file",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
263 dest="input_fasta",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
264 type="string"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
265 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
266 parser.add_option(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
267 "--output",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
268 dest="output",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
269 type="string"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
270 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
271 parser.add_option(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
272 "--plot-output",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
273 type="string",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
274 dest="plot_output"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
275 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
276 parser.add_option(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
277 "--json",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
278 dest="json_output",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
279 type="string"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
280 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
281 parser.add_option(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
282 "--plot",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
283 action="store_true"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
284 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
285 parser.add_option(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
286 "--plot_all",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
287 action="store_true"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
288 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
289 parser.add_option(
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
290 "--highlight",
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
291 action="store_true"
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
292 )
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
293 try:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
294 options, args = parser.parse_args()
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
295 if not (options.dynamine or options.disomine or options.efoldmine or options.agmata):
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
296 parser.error('At least one predictor is required')
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
297 if not options.input_fasta:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
298 parser.error('Input file not given (--file)')
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
299 if not options.output:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
300 parser.error('Output directory not given (--output)')
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
301 if (options.plot or options.plot_all) and not options.plot_output:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
302 parser.error('Plot output directory not given (--plot-output)')
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
303 if not options.json_output:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
304 parser.error('Json output file not given (--json)')
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
305 main(options)
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
306 except optparse.OptionError as exc:
dc91f5626375 planemo upload commit 599e1135baba020195b3f7576449d595bca9af75
iuc
parents:
diff changeset
307 raise RuntimeError(f"Invalid arguments: {args}") from exc