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
|
"""module contains pca implementation using python"""
from typing import Any
from scipy import stats
from sklearn.decomposition import PCA
from sklearn import preprocessing
import numpy as np
import redis
from typing_extensions import TypeAlias
fArray: TypeAlias = list[float]
def compute_pca(array: list[fArray]) -> dict[str, Any]:
"""
computes the principal component analysis
Parameters:
array(list[list]):a list of lists contains data to perform pca
Returns:
pca_dict(dict):dict contains the pca_object,pca components,pca scores
"""
corr_matrix = np.array(array)
pca_obj = PCA()
scaled_data = preprocessing.scale(corr_matrix)
pca_obj.fit(scaled_data)
return {
"pca": pca_obj,
"components": pca_obj.components_,
"scores": pca_obj.transform(scaled_data)
}
def generate_scree_plot_data(variance_ratio: fArray) -> tuple[list, fArray]:
"""
generates the scree data for plotting
Parameters:
variance_ratio(list[floats]):ratios for contribution of each pca
Returns:
coordinates(list[(x_coor,y_coord)])
"""
perc_var = [round(ratio*100, 1) for ratio in variance_ratio]
x_coordinates = [f"PC{val}" for val in range(1, len(perc_var)+1)]
return (x_coordinates, perc_var)
def generate_pca_traits_vals(trait_data_array: list[fArray],
corr_array: list[fArray]) -> list[list[Any]]:
"""
generates datasets from zscores of the traits and eigen_vectors\
of correlation matrix
Parameters:
trait_data_array(list[floats]):an list of the traits
corr_array(list[list]): list of arrays for computing eigen_vectors
Returns:
pca_vals[list[list]]:
"""
trait_zscores = stats.zscore(trait_data_array)
if len(trait_data_array[0]) < 10:
trait_zscores = trait_data_array
(eigen_values, corr_eigen_vectors) = np.linalg.eig(np.array(corr_array))
idx = eigen_values.argsort()[::-1]
return np.dot(corr_eigen_vectors[:, idx], trait_zscores)
def process_factor_loadings_tdata(factor_loadings, traits_num: int):
"""
transform loadings for tables visualization
Parameters:
factor_loading(numpy.ndarray)
traits_num(int):number of traits
Returns:
tabular_loadings(list[list[float]])
"""
target_columns = 3 if traits_num > 2 else 2
trait_loadings = list(factor_loadings.T)
return [list(trait_loading[:target_columns])
for trait_loading in trait_loadings]
def generate_pca_temp_traits(
species: str,
group: str,
traits_data: list[fArray],
corr_array: list[fArray],
dataset_samples: list[str],
shared_samples: list[str],
create_time: str
) -> dict[str, list[Any]]:
"""
generate pca temp datasets
"""
# pylint: disable=too-many-arguments
pca_trait_dict = {}
pca_vals = generate_pca_traits_vals(traits_data, corr_array)
for (idx, pca_trait) in enumerate(list(pca_vals)):
trait_id = f"PCA{str(idx+1)}_{species}_{group}_{create_time}"
sample_vals = []
pointer = 0
for sample in dataset_samples:
if sample in shared_samples:
sample_vals.append(str(pca_trait[pointer]))
pointer += 1
else:
sample_vals.append("x")
pca_trait_dict[trait_id] = sample_vals
return pca_trait_dict
def cache_pca_dataset(redis_conn: Any, exp_days: int,
pca_trait_dict: dict[str, list[Any]]):
"""
caches pca dataset to redis
Parameters:
redis_conn(object)
exp_days(int): fo redis cache
pca_trait_dict(Dict): contains traits and traits vals to cache
Returns:
boolean(True if correct conn object False incase of exception)
"""
try:
for trait_id, sample_data in pca_trait_dict.items():
samples_str = " ".join([str(x) for x in sample_data])
redis_conn.set(trait_id, samples_str, ex=exp_days)
return True
except (redis.ConnectionError, AttributeError):
return False
|