-
Notifications
You must be signed in to change notification settings - Fork 1
/
PLSR_phospho_metabo.py
518 lines (388 loc) · 19.3 KB
/
PLSR_phospho_metabo.py
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
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
# -*- coding: utf-8 -*-
"""
@authors: TYT, RSM, FV
original code developed by RSM can be accessed via: https://github.com/RackS103/Omics_Analysis_Scripts
@machine: Luisa
"""
import os.path
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from scipy import stats
import seaborn as sns
from scipy.signal import savgol_filter
from sys import stdout
# from Enrichment_Scripts_fromRM import run_Enrichr, run_KEA3, run_STRING
# from PLSDA_fromRM import PLSClassifier
from mbpls.mbpls import MBPLS
from sklearn.cross_decomposition import PLSRegression
from sklearn.metrics import mean_squared_error, mean_absolute_error, accuracy_score, r2_score, roc_auc_score, roc_curve, RocCurveDisplay
from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.impute import KNNImputer
from sklearn.ensemble import RandomForestClassifier
from sklearn.decomposition import PCA
from sklearn.base import TransformerMixin, ClassifierMixin
from sklearn.model_selection import GridSearchCV, LeaveOneOut, KFold, train_test_split, cross_val_predict
from sklearn.pipeline import Pipeline
from sklearn import metrics
from natsort import natsorted
from scipy.stats import ttest_ind
def vip(model, feature_labels, X):
"""
Regex_1 and Regex_2 is specific to Tig's project and is used to add in the Fold changes for two populations
model <- PLSR model
feature_labels <- columns of
X <- X phospho matrix
"""
t = model.x_scores_
w = model.x_weights_
q = model.y_loadings_
p, h = w.shape
vips = np.zeros((p,))
foldsF = np.zeros((p,))
foldsM = np.zeros((p,))
df=X.T
FNCD = df.loc[:, [int(col[1:]) <= 19 and col in df.columns[df.columns.str.contains('|'.join(['F']))==True] for col in df.columns] ]
FHFD=df.loc[:, [int(col[1:]) > 19 and col in df.columns[df.columns.str.contains('|'.join(['F']))==True] for col in df.columns] ]
MNCD=df.loc[:, [int(col[1:]) <= 20 and col in df.columns[df.columns.str.contains('|'.join(['M']))==True] for col in df.columns] ]
MHFD=df.loc[:, [int(col[1:]) > 20 and col in df.columns[df.columns.str.contains('|'.join(['M']))==True] for col in df.columns] ]
s = np.diag(t.T @ t @ q.T @ q).reshape(h, -1)
total_s = np.sum(s)
for i in range(p):
weight = np.array([ (w[i,j] / np.linalg.norm(w[:,j]))**2 for j in range(h) ])
vips[i] = np.sqrt(p*(s.T @ weight)/total_s)
foldsF[i] = np.mean(FHFD.T.iloc[:, i]) / np.mean(FNCD.T.iloc[:, i])
foldsM[i] = np.mean(MHFD.T.iloc[:, i]) / np.mean(MNCD.T.iloc[:, i])
coef_col = 0
vips =pd.concat([pd.DataFrame({'VIP': vips, 'Coef':model.coef_[:,coef_col],
'F_Log2(HFD/NCD)': np.log2(foldsF), 'M_Log2(HFD/NCD)': np.log2(foldsM) }, index = feature_labels),df], axis=1)
vips=vips.sort_values(by='VIP', ascending=False)
return vips
#this is for a single Y variable
def feature_selection(X, y, y_name, n):
'''
Selects the top N (default 50) features with the highest magnitude correlation to the given Y variable.
Parameters:
X (DataFrame): Feature matrix.
y (Series): Target variable.
y_name (str): Name of the target variable.
n (int): Number of top features to select.
Returns:
DataFrame: Selected features.
'''
# Calculate correlation coefficients between each feature and the target variable
corr_values = X.apply(lambda col: np.corrcoef(col, y)[0, 1])
# Create DataFrame to store correlation coefficients
corr_df = pd.DataFrame({'Feature': X.columns, 'Correlation': corr_values.abs()})
# Select top N features based on absolute correlation values
selected_features = corr_df.nlargest(n, 'Correlation')['Feature']
# Return selected features
return X[selected_features]
def loocv_score_singleY(model, scorer, X, Y):
"""
Q^2 score for univariate Y matrix.
model <- sklearn model
scorer <- scoring function
X <- X matrix, pandas format only
Y <- Y matrix, should be a 1D vector/pandas Series.
"""
loo = LeaveOneOut()
Y_hat_test = np.zeros(Y.shape)
train_scores = []
latent_variable_scores = []
percent_vars_explained = []
loadings = []
for train_idx, test_idx in loo.split(X):
X_train = X[train_idx, :]
X_test = X[test_idx, :]
Y_train = Y[train_idx]
scaler = StandardScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)
model.fit(X_train, Y_train)
Y_hat_train = model.predict(X_train)
train_scores.append(scorer(Y_train, Y_hat_train))
Y_hat_test[test_idx] = model.predict(X_test)
# Extract latent variable scores
latent_variable_scores.append(model.transform(X_test)[:, :2]) # Assuming 2 latent variables
# Calculate variance explained for the first two components
u, s, _ = np.linalg.svd(X_train, full_matrices=False)
explained_variance = (s**2) / np.sum(s**2)
percent_vars_explained.append(explained_variance[:2] * 100) # First two components
# Extract loadings
loadings.append(model.x_loadings_[:, :2]) # Assuming 2 latent variables
# Calculate Q²
press = np.sum((Y - Y_hat_test) ** 2)
tss = np.sum((Y - np.mean(Y)) ** 2)
q2 = 1 - press / tss
# Calculate RMSE and MAE
rmse = mean_squared_error(Y, Y_hat_test, squared=False)
mae = mean_absolute_error(Y, Y_hat_test)
# Create a dataframe for the results
LV_df = pd.DataFrame({
'Percent Variance Explained': percent_vars_explained,
'Latent Variable Scores': latent_variable_scores,
'Loadings': loadings
})
return np.mean(train_scores), scorer(Y, Y_hat_test), LV_df, q2, rmse, mae
def PLS_CV(X, Y, model_class=PLSRegression, gs_scoring='neg_mean_squared_error', score_fx=r2_score, cv_range=np.arange(2,20,2), multi_Y=False, verbose=False):
"""
X <- phospho matrix
Y <- Metabolomics matrix
model_class <- type of model to use (PLSRegression or PLSClassifier)
gs_scoring <- sklearn scoring function string name, used in GridSearchCV to find optimal n_components for model
score_fx <- function to score the performance of model.
cv_range <- range of values to test for n_components in PLSR
multi_Y <- True if Y matrix is multivariate
verbose <- prints out results as function works if True
"""
if multi_Y:
Y = StandardScaler().fit_transform(Y)
else:
Y = stats.zscore(Y.astype(float))
print(model_class())
pipe = Pipeline(steps=[('scaler', StandardScaler()), ('predictor', model_class())])
gs = GridSearchCV(estimator=pipe, param_grid={'predictor__n_components':cv_range},
cv=LeaveOneOut(), scoring=gs_scoring)
gs.fit(X, Y)
ncomp = gs.best_params_['predictor__n_components']
model = model_class(n_components=ncomp)
model.fit(X, Y)
y_c = model.predict(X)
# Cross-val
y_cv = cross_val_predict(model, X, Y, cv=LeaveOneOut())
# Calibration scores and errors
r2_c = r2_score(Y, y_c)
rmse_c = mean_squared_error(Y, y_c, squared=False)
mae_c = mean_absolute_error(Y, y_c)
# Cross-val scores and errors
r2_cv = r2_score(Y, y_cv)
rmse_cv = mean_squared_error(Y, y_cv, squared=False)
mae_cv = mean_absolute_error(Y, y_cv)
if verbose:
print(f'Best model was {model} with {gs_scoring} {gs.best_score_}')
print(f'Calibration R²: {r2_c}, Calibration RMSE: {rmse_c}, Calibration MAE: {mae_c}')
print(f'Cross-validated R²: {r2_cv}, Cross-validated RMSE: {rmse_cv}, Cross-validated MAE: {mae_cv}')
if multi_Y:
train_score, test_score = loocv_score_multiY(model, score_fx, X.to_numpy(), Y)
else:
train_score, test_score, LVs, q2, rmse, mae = loocv_score_singleY(model, score_fx, X.to_numpy(), Y)
if verbose:
print(f'Train Performance ({score_fx.__name__}): {train_score}')
print(f'Test Performance ({score_fx.__name__}): {test_score}')
print(f'Predictive Ability (Q²): {q2}')
print(f'Cross-validated RMSE: {rmse}')
print(f'Cross-validated MAE: {mae}')
return model, r2_c, rmse_c, mae_c, q2, rmse_cv, mae_cv,ncomp, LVs
def kfold_score_singleY(model, scorer, X, Y):
"""
Q^2 score for univariate Y matrix.
model <- sklearn model
scorer <- scoring function
X <- X matrix, pandas format only
Y <- Y matrix, should be a 1D vector/pandas Series.
"""
loo = KFold(n_splits=7)
Y_hat_test = np.zeros(Y.shape)
train_scores = []
latent_variable_scores = []
percent_vars_explained = []
loadings = []
for train_idx, test_idx in loo.split(X):
X_train = X[train_idx, :]
X_test = X[test_idx, :]
Y_train = Y[train_idx]
scaler = StandardScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)
model.fit(X_train, Y_train)
Y_hat_train = model.predict(X_train)
train_scores.append(scorer(Y_train, Y_hat_train))
# Ensure predictions are 1D
Y_hat_test[test_idx] = model.predict(X_test).flatten()
# Extract latent variable scores
latent_variable_scores.append(model.transform(X_test)[:, :2]) # Assuming 2 latent variables
# Calculate variance explained for the first two components
u, s, _ = np.linalg.svd(X_train, full_matrices=False)
explained_variance = (s**2) / np.sum(s**2)
percent_vars_explained.append(explained_variance[:2] * 100) # First two components
# Extract loadings
loadings.append(model.x_loadings_[:, :2]) # Assuming 2 latent variables
# Calculate Q²
press = np.sum((Y - Y_hat_test) ** 2)
tss = np.sum((Y - np.mean(Y)) ** 2)
q2 = 1 - press / tss
# Calculate RMSE and MAE
rmse = mean_squared_error(Y, Y_hat_test, squared=False)
mae = mean_absolute_error(Y, Y_hat_test)
# Create a dataframe for the results
LV_df = pd.DataFrame({
'Percent Variance Explained': percent_vars_explained,
'Latent Variable Scores': latent_variable_scores,
'Loadings': loadings
})
return np.mean(train_scores), scorer(Y, Y_hat_test), LV_df, q2, rmse, mae
def kfold_PLS_CV(X, Y, model_class=PLSRegression, gs_scoring='neg_mean_squared_error', score_fx=r2_score, cv_range=np.arange(1, 40, 1), multi_Y=False, verbose=False):
"""
X <- phospho matrix
Y <- Metabolomics matrix
model_class <- type of model to use (PLSRegression or PLSClassifier)
gs_scoring <- sklearn scoring function string name, used in GridSearchCV to find optimal n_components for model
score_fx <- function to score the performance of model.
cv_range <- range of values to test for n_components in PLSR
multi_Y <- True if Y matrix is multivariate
verbose <- prints out results as function works if True
"""
if multi_Y:
Y = StandardScaler().fit_transform(Y)
else:
Y = stats.zscore(Y.astype(float))
pipe = Pipeline(steps=[('scaler', StandardScaler()), ('predictor', model_class())])
# First stage grid search
gs = GridSearchCV(estimator=pipe, param_grid={'predictor__n_components':cv_range},
cv=KFold(n_splits=7), scoring=gs_scoring)
gs.fit(X, Y)
# Refine around the best parameter found in the first stage
best_ncomp = gs.best_params_['predictor__n_components']
refined_range = np.arange(max(1, best_ncomp-5), best_ncomp+6)
gs_refined = GridSearchCV(estimator=pipe, param_grid={'predictor__n_components':refined_range},
cv=KFold(n_splits=7), scoring=gs_scoring)
gs_refined.fit(X, Y)
ncomp = gs_refined.best_params_['predictor__n_components']
model = model_class(n_components=ncomp)
model.fit(X, Y)
y_c = model.predict(X)
# Cross-val
y_cv = cross_val_predict(model, X, Y, cv=KFold(n_splits=7))
# Calibration scores and errors
r2_c = r2_score(Y, y_c)
rmse_c = mean_squared_error(Y, y_c, squared=False)
mae_c = mean_absolute_error(Y, y_c)
# Cross-val scores and errors
r2_cv = r2_score(Y, y_cv)
rmse_cv = mean_squared_error(Y, y_cv, squared=False)
mae_cv = mean_absolute_error(Y, y_cv)
if verbose:
print(f'Best model was {model} with {gs_scoring} {gs.best_score_}')
print(f'Calibration R²: {r2_c}, Calibration RMSE: {rmse_c}, Calibration MAE: {mae_c}')
print(f'Cross-validated R²: {r2_cv}, Cross-validated RMSE: {rmse_cv}, Cross-validated MAE: {mae_cv}')
if multi_Y:
train_score, test_score = loocv_score_multiY(model, score_fx, X.to_numpy(), Y)
else:
train_score, test_score, LVs, q2, rmse, mae = loocv_score_singleY(model, score_fx, X.to_numpy(), Y)
if verbose:
print(f'Train Performance ({score_fx.__name__}): {train_score}')
print(f'Test Performance ({score_fx.__name__}): {test_score}')
print(f'Predictive Ability (Q²): {q2}')
print(f'Cross-validated RMSE: {rmse}')
print(f'Cross-validated MAE: {mae}')
return model, r2_c, rmse_c, mae_c, q2, rmse_cv, mae_cv,ncomp, LVs
def do_analysis(X, Y, name, expType, classifier):
model, r2_c, rmse_c, mae_c, q2, rmse_cv, mae_cv,ncomp, LVs= PLS_CV(X,Y)
newData['Name'].append(name)
newData['Q2'].append(q2)
newData['R2'].append(r2_c)
newData['RMSE_cv'].append(rmse_cv)
newData['MAE_cv'].append(mae_cv)
newData['RMSE_training'].append(rmse_c)
newData['MAE_training'].append(mae_c)
newData['Model'].append(model)
newData['Components'].append(ncomp)
if q2>=0.4:
t = model.x_scores_
w = model.x_weights_
q = model.y_loadings_
p, h = w.shape
vip = np.zeros((p,))
foldsF = np.zeros((p,))
foldsM = np.zeros((p,))
pvalsF = np.zeros((p,))
pvalsM = np.zeros((p,))
df=X.T
if "BHA" in expType:
FNCD = df.loc[:, [int(col[2:]) <= 5 and col in df.columns[df.columns.str.contains('|'.join(['F']))==True] for col in df.columns] ].T
FHFD=df.loc[:, [int(col[2:]) > 5 and col in df.columns[df.columns.str.contains('|'.join(['F']))==True] for col in df.columns] ].T
MNCD=df.loc[:, [int(col[2:]) <= 10 and col in df.columns[df.columns.str.contains('|'.join(['M']))==True] for col in df.columns] ].T
MHFD=df.loc[:, [int(col[2:]) > 10 and col in df.columns[df.columns.str.contains('|'.join(['M']))==True] for col in df.columns] ].T
else:
FNCD = df.loc[:, [int(col[1:]) <= 19 and col in df.columns[df.columns.str.contains('|'.join(['F']))==True] for col in df.columns] ].T
FHFD=df.loc[:, [int(col[1:]) > 19 and col in df.columns[df.columns.str.contains('|'.join(['F']))==True] for col in df.columns] ].T
MNCD=df.loc[:, [int(col[1:]) <= 20 and col in df.columns[df.columns.str.contains('|'.join(['M']))==True] for col in df.columns] ].T
MHFD=df.loc[:, [int(col[1:]) > 20 and col in df.columns[df.columns.str.contains('|'.join(['M']))==True] for col in df.columns] ].T
s = np.diag(t.T @ t @ q.T @ q).reshape(h, -1)
total_s = np.sum(s)
for i in range(p):
weight = np.array([ (w[i,j] / np.linalg.norm(w[:,j]))**2 for j in range(h) ])
vip[i] = np.sqrt(p*(s.T @ weight)/total_s)
foldsF[i] = np.mean(FHFD.iloc[:, i]) / np.mean(FNCD.iloc[:, i])
t, pvalsF[i] = ttest_ind(FNCD.iloc[:,i], FHFD.iloc[:,i])
foldsM[i] = np.mean(MHFD.iloc[:, i]) / np.mean(MNCD.iloc[:, i])
t, pvalsM[i] = ttest_ind(MNCD.iloc[:,i], MHFD.iloc[:,i])
coef_col = 0
vips[name] = pd.concat([pd.DataFrame({'VIP': vip, 'Coef':model.coef_[:,coef_col],'F_Log2(HFD/NCD)': np.log2(foldsF), 'M_Log2(HFD/NCD)': np.log2(foldsM) }, index = X.columns),df], axis=1)
return newData, vips, LVs
# read in all phospho and metabo data for first chorot: HFD v NCD
Xm = pd.read_csv('NCDvHFD_70mc-Metabolomics.csv', index_col='Metabolites')
Xm = Xm.fillna(1).T
Xm_f=Xm.iloc[0:38,:]
Xm_m=Xm.iloc[38:,:]
Xy= pd.read_csv('NCDvHFD70_PRM_new_clusterTable.csv', index_col='Label')
Xy=Xy.drop(axis=1, labels=['Clusters'])
Xy=Xy.reindex(columns=natsorted(Xy.columns))
Xy= Xy.T
Xy_f=Xy.iloc[0:38,:]
Xy_m=Xy.iloc[38:,:]
#read in all phospho & metabo data for BHA chorot: HFD v HFD+BHA
Xm = pd.read_csv('20240429_BHA_70.csv', index_col='Class_Metabolites').drop(['Class','Metabolites'], axis=1)
Xm = Xm.fillna(0).T
Xm_f=Xm.iloc[0:10,:]
Xm_m=Xm.iloc[10:,:]
Xy=pd.read_csv('BHA_PRM_pY_brgA_u_wClass.csv', index_col=['Label']).drop(['extras','Structural_annot','Class', 'Sites'], axis=1)
Xy=Xy.fillna(1)
Xy= Xy.T
Xy_f=Xy.iloc[0:10,:]
Xy_m=Xy.iloc[10:,:]
# Data dictionary
newData = {'Name':[], 'Q2':[],'R2':[],'RMSE_cv':[],'MAE_cv':[],'RMSE_training':[],'MAE_training':[], 'Model':[], 'Components':[]}
vips={}
LVs_dict={}
# Perform PLSR on single Y with or without feature selection
for i in range(len(Xm_f.columns)):
expType="BHA"
phenos = Xm_f.columns
name = phenos[i]
# Identify common rows
common_index = Xm_f.index.intersection(Xy_f.index)
# Select features present in both dataframes
X_sel = Xy_f.loc[common_index]
# Select features present in both dataframes
y= Xm_f.loc[common_index]
# Perform feature selection: if running with no FS then use n=len(X_sel.T), for top 50% of co-correlated features use n=len(X_sel.T)//2
X_sel = feature_selection(X_sel, y.loc[common_index, name], name, n=len(X_sel.T)//2)
# Perform analysis with selected features
newData, vips, LVs_dict[name] = do_analysis(X_sel,y[name], name, expType, classifier=False)
# Create DataFrame from selected keys
newDataframe = pd.DataFrame({key: newData[key] for key in ['Name', 'Q2','R2','RMSE_cv','MAE_cv', 'RMSE_training','MAE_training', 'Model', 'Components']})
newDataframe=newDataframe.sort_values(by='Q2',ascending=False)
newDataframe.to_csv("F_PRM_BHA_top50p_PLSR.csv")
PLSRhits=newDataframe.loc[newDataframe['Q2']>=0.4, 'Name'].tolist()
directory_name = "F_PRM_BHA_top50p_VIPs"
if not os.path.exists(directory_name):
os.mkdir(directory_name)
for k, v in vips.items():
if k in PLSRhits:
k=k.replace('/','_')
v=v.sort_values(by='VIP', ascending=False)
filename = f"{k}_VIPs.csv"
filepath = os.path.join(directory_name, filename)
v.to_csv(filepath, index=True)
directory_name = "F_PRM_BHA_top50p_LVs"
if not os.path.exists(directory_name):
os.mkdir(directory_name)
for k, v in LVs_dict.items():
if k in PLSRhits:
k=k.replace('/','_')
filename = f"{k}_LVs.csv"
filepath = os.path.join(directory_name, filename)
v.to_csv(filepath, index=True)