-
Notifications
You must be signed in to change notification settings - Fork 0
/
ALL_imputation_chunk.py
304 lines (240 loc) · 15.2 KB
/
ALL_imputation_chunk.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
import numpy as np
from sklearn.svm import SVR
import pandas as pd
from sklearn.model_selection import cross_val_score
from statistics import mean
import math
from sklearn.linear_model import LinearRegression
from sklearn.preprocessing import scale
from pandas import DataFrame
import pickle
from sklearn.ensemble import RandomForestRegressor
from sklearn.neighbors import KNeighborsRegressor
import time
from scipy import stats
import argparse
from sklearn.linear_model import ElasticNet
from sklearn.model_selection import GridSearchCV
from sklearn.metrics import mean_squared_error
from sklearn.metrics import explained_variance_score
from sklearn.metrics import r2_score
from sklearn.metrics import make_scorer#use to convert metrics to scoring callables
mse = make_scorer(mean_squared_error, greater_is_better=False)
r2 = make_scorer(r2_score, greater_is_better=True)
evs = make_scorer(explained_variance_score, greater_is_better=True)
parser = argparse.ArgumentParser()
parser.add_argument("chr", action="store", help="put chromosome no")
parser.add_argument("chunk", action="store", help="put chromosome chunk no")
args = parser.parse_args()
chrom = str(args.chr)
chunk = str(args.chunk)
tr_pop = "ALL"
pop = "CAU_thrombomodulin_rankplt5_pheno"
#important functions needed
def get_filtered_snp_annot (snpfilepath):
snpanot = pd.read_csv(snpfilepath, sep="\t")
snpanot = snpanot[(((snpanot["refAllele"]=="A") & (snpanot["effectAllele"]=="C")) | ((snpanot["refAllele"]=="C") & (snpanot["effectAllele"]=="A")) | ((snpanot["refAllele"]=="A") & (snpanot["effectAllele"]=="G")) | ((snpanot["refAllele"]=="G") & (snpanot["effectAllele"]=="A")) | ((snpanot["refAllele"]=="T") & (snpanot["effectAllele"]=="G")) | ((snpanot["refAllele"]=="G") & (snpanot["effectAllele"]=="T")) | ((snpanot["refAllele"]=="T") & (snpanot["effectAllele"]=="C")) | ((snpanot["refAllele"]=="C") & (snpanot["effectAllele"]=="T"))) & (snpanot["rsid"].notna())]
snpanot = snpanot.drop_duplicates(["varID"])
return snpanot
def get_gene_annotation (gene_anot_filepath, chrom, gene_types=["protein_coding"]):
gene_anot = pd.read_csv(gene_anot_filepath, sep="\t")
gene_anot = gene_anot[(gene_anot["chr"]==str(chrom)) & (gene_anot["gene_type"].isin(gene_types))]
return gene_anot
def get_gene_type (gene_anot, gene):
gene_type = gene_anot[gene_anot["gene_id"]==gene]
gene_type = gene_type.iloc[0,5]
return gene_type
def get_gene_name (gene_anot, gene):
gene_name = gene_anot[gene_anot["gene_id"]==gene]
gene_name = gene_name.iloc[0,2]
return gene_name
def get_gene_coords (gene_anot, gene):
gene_type = gene_anot[gene_anot["gene_id"]==gene]
gene_coord = [gene_type.iloc[0,3], gene_type.iloc[0,4]]
return gene_coord
def get_covariates (cov_filepath):
cov = pd.read_csv(cov_filepath, sep=" ")
cov = cov.set_index("IID") #make IID to be the row names
cov.index.names = [None] # remove the iid name from the row
pc = ["PC1", "PC2", "PC3"] #a list of the PCs to retain
cov = cov[pc]
return cov
def get_gene_expression(gene_expression_file_name, gene_annot):
expr_df = pd.read_csv(gene_expression_file_name, header = 0, index_col = 0, delimiter='\t')
expr_df = expr_df.T
inter = list(set(gene_annot['gene_id']).intersection(set(expr_df.columns)))
#print(len(inter))
expr_df = expr_df.loc[:, inter ]
return expr_df
def adjust_for_covariates (expr_vec, cov_df):
reg = LinearRegression().fit(cov_df, expr_vec)
ypred = reg.predict(cov_df)
residuals = expr_vec - ypred
residuals = scale(residuals)
return residuals
def get_maf_filtered_genotype(genotype_file_name, maf):
gt_df = pd.read_csv(genotype_file_name, 'r', header = 0, index_col = 0,delimiter='\t')
effect_allele_freqs = gt_df.mean(axis=1)
effect_allele_freqs = [ x / 2 for x in effect_allele_freqs ]
effect_allele_boolean = pd.Series([ ((x >= maf) & (x <= (1 - maf))) for x in effect_allele_freqs ]).values
gt_df = gt_df.loc[ effect_allele_boolean ]
gt_df = gt_df.T
return gt_df
def get_cis_genotype (gt_df, snp_annot, coords, cis_window=1000000):
snp_info = snpannot[(snpannot['pos'] >= (coords[0] - cis_window)) & (snpannot['rsid'].notna()) & (snpannot['pos'] <= (coords[1] + cis_window))]
if len(snp_info) == 0:
return 0
else:
gtdf_col = list(gt_df.columns)
snpinfo_col = list(snp_info["varID"])
intersect = snps_intersect(gtdf_col, snpinfo_col) #this function was defined earlier
cis_gt = gt_df[intersect]
return cis_gt
def calc_R2 (y, y_pred):
tss = 0
rss = 0
for i in range(len(y)):
tss = tss + (y[i])**2
rss = rss + (((y[i]) - (y_pred[i]))**2)
tss = float(tss)
rss = float(rss)
r2 = 1 - (rss/tss)
return r2
def calc_corr (y, y_pred):
num = 0
dem1 = 0
dem2 = 0
for i in range(len(y)):
num = num + ((y[i]) * (y_pred[i]))
dem1 = dem1 + (y[i])**2
dem2 = dem2 + (y_pred[i])**2
num = float(num)
dem1 = math.sqrt(float(dem1))
dem2 = math.sqrt(float(dem2))
rho = num/(dem1*dem2)
return rho
def snps_intersect(list1, list2):
return list(set(list1) & set(list2))
tr_snp = "/home/rschubert1/data/split_genotypes_for_paul/"+tr_pop+"/sliced_genotypes/"+tr_pop+"_chr"+chrom+"_genotype_chunk"+chunk+".txt.gz"
gex = "/home/rschubert1/data/split_genotypes_for_paul/"+tr_pop+"/chunked_expression/"+tr_pop+"_chr"+chrom+"_gex_chunk"+chunk+".txt.gz"
cov_file = "/home/rschubert1/data/split_genotypes_for_paul/covariates/PC3_"+tr_pop+"_PCs_sorted.txt"
geneanotfile = "/home/pokoro/data/mesa_models/gencode.v18.annotation.parsed.txt"
snpfilepath = "/home/rschubert1/data/split_genotypes_for_paul/anno/"+tr_pop+".chr"+chrom+".anno.txt.gz"
#train functioning
snpannot = get_filtered_snp_annot(snpfilepath)
geneannot = get_gene_annotation(geneanotfile, chrom)
cov = get_covariates(cov_file)
expr_df = get_gene_expression(gex, geneannot)
expr_df.drop(axis=0, inplace=True,labels="PROBE_ID") #Remove the PROBE_ID because ryan did not remove it when he created the expression files, and it causes error downstream
genes = list(expr_df.columns)
gt_df = get_maf_filtered_genotype(tr_snp, 0.01)
#test data files
test_snp = "/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/cau_imputation_dosage_chr"+chrom+"_chunk"+chunk+".txt"
test_annot = "/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/cau_imputation_dosage_chr"+chrom+"_annot.txt"
#test functioning
test_snpannot = get_filtered_snp_annot(test_annot)
test_gt_df = get_maf_filtered_genotype(test_snp, 0.01)
test_ids = list(test_gt_df.index)
#frame to store the ypred and test adjusted expression
ypred_frame_rf = pd.DataFrame()
ypred_frame_svr = pd.DataFrame()
ypred_frame_knn = pd.DataFrame()
#read in the grid search best result files and take the params to fit the model
rf_grid = pd.read_csv("/home/pokoro/data/mesa_models/python_ml_models/ALL_results/grid_split/"+tr_pop+"_best_grid_split_rf_cv_chr"+chrom+"_chunk"+chunk+".txt", sep="\t")
knn_grid = pd.read_csv("/home/pokoro/data/mesa_models/python_ml_models/ALL_results/grid_split/"+tr_pop+"_best_grid_split_knn_cv_chr"+chrom+"_chunk"+chunk+".txt", sep="\t")
svr_grid = pd.read_csv("/home/pokoro/data/mesa_models/python_ml_models/ALL_results/grid_split/"+tr_pop+"_best_grid_split_svr_cv_chr"+chrom+"_chunk"+chunk+".txt", sep="\t")
#create file with header to write out expression to file immediately
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"rf_pred_expr.txt", "a").write("gene_id")
for i in range(len(test_ids)):
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"rf_pred_expr.txt", "a").write("\t" + str(test_ids[i]))
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"knn_pred_expr.txt", "a").write("gene_id")
for i in range(len(test_ids)):
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"knn_pred_expr.txt", "a").write("\t" + str(test_ids[i]))
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"svr_pred_expr.txt", "a").write("gene_id")
for i in range(len(test_ids)):
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"svr_pred_expr.txt", "a").write("\t" + str(test_ids[i]))
#algorithms to use
for gene in genes:
coords = get_gene_coords(geneannot, gene)
gene_name = get_gene_name(geneannot, gene)
#print(gene)
expr_vec = expr_df[gene]#observed exp
#print(expr_vec)
adj_exp = adjust_for_covariates(list(expr_vec), cov)#adjusted exp
cis_gt = get_cis_genotype(gt_df, snpannot, coords)
test_cis_gt = get_cis_genotype(test_gt_df, test_snpannot, coords)
if (type(cis_gt) != int) & (type(test_cis_gt) != int):#just to be sure the cis genotype is not empty
gg = [gene] #just to cast the gene id to list because pandas need it to be in list before it can be used as col name
#take the snps
train_snps = list(cis_gt.columns)
test_snps = list(test_cis_gt.columns)
snp_intersect = snps_intersect(train_snps, test_snps)
cis_gt = cis_gt[snp_intersect]
test_cis_gt = test_cis_gt[snp_intersect]
if (cis_gt.shape[1] > 0) & (test_cis_gt.shape[1] > 0): #make sure that the cis_gt is not empty
#build the model
cis_gt = cis_gt.values
test_cis_gt = test_cis_gt.values
#Random Forest
gnlist = list(rf_grid['Gene_Name'])
f = gene_name in gnlist
if f != False: #Just to be sure that the gene exist in the RF best grid dataframe
n_tree = rf_grid[rf_grid['Gene_Name']==gene_name].iloc[0,3]
rf = RandomForestRegressor(random_state=1234, n_estimators=n_tree)
rf.fit(cis_gt, adj_exp.ravel())
ypred = rf.predict(test_cis_gt)
#write out ypred quickly
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"rf_pred_expr.txt", "a").write("\n")
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"rf_pred_expr.txt", "a").write(str(gene))
for j in range(len(ypred)):
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"rf_pred_expr.txt", "a").write("\t"+str(ypred[j]))
#prepare ypred for writing out to a file
ypred_pd = pd.DataFrame(ypred)
ypred_pd.columns = gg
ypred_pd.index = test_ids
ypred_frame_rf = pd.concat([ypred_frame_rf, ypred_pd], axis=1, sort=True)
#Support Vector Machine
gnlist = list(svr_grid['Gene_Name'])
f = gene_name in gnlist
if f != False: #Just to be sure that the gene exist in the SVR best grid dataframe
kernel = svr_grid[svr_grid['Gene_Name']==gene_name].iloc[0,3]
degree = svr_grid[svr_grid['Gene_Name']==gene_name].iloc[0,4]
c = svr_grid[svr_grid['Gene_Name']==gene_name].iloc[0,5]
svr = SVR(gamma="scale", kernel=kernel, degree=degree, C=c)
svr.fit(cis_gt, adj_exp.ravel())
ypred = svr.predict(test_cis_gt)
#write out ypred quickly
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"svr_pred_expr.txt", "a").write("\n")
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"svr_pred_expr.txt", "a").write(str(gene))
for j in range(len(ypred)):
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"svr_pred_expr.txt", "a").write("\t"+str(ypred[j]))
#prepare ypred for writing out to a file
yprep_pd = pd.DataFrame(ypred)
ypred_pd.columns = gg
ypred_pd.index = test_ids
ypred_frame_svr = pd.concat([ypred_frame_svr, ypred_pd], axis=1, sort=True)
#K Nearest Neighbour
gnlist = list(knn_grid['Gene_Name'])
f = gene_name in gnlist
if f != False: #Just to be sure that the gene exist in the KNN best grid dataframe
k = knn_grid[knn_grid['Gene_Name']==gene_name].iloc[0,3]
weight = knn_grid[knn_grid['Gene_Name']==gene_name].iloc[0,4]
knn = KNeighborsRegressor(n_neighbors=k, weights = weight)
knn.fit(cis_gt, adj_exp.ravel())
ypred = knn.predict(test_cis_gt)
#write out ypred quickly
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"knn_pred_expr.txt", "a").write("\n")
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"knn_pred_expr.txt", "a").write(str(gene))
for j in range(len(ypred)):
open("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"knn_pred_expr.txt", "a").write("\t"+str(ypred[j]))
#prepare ypred for writing out to a file
yprep_pd = pd.DataFrame(ypred)
ypred_pd.columns = gg
ypred_pd.index = test_ids
ypred_frame_knn = pd.concat([ypred_frame_knn, ypred_pd], axis=1, sort=True)
#There was error writing out to this file below because i failed to include the chunk before. so delete all these files in the result chunk folder for chrom 1-8, 22,20,14,19,17,16,15
#delete only there rf. eg is chr8_chunk1_ALL_2_CAU_thrombomodulin_rankplt5_pheno_rf_pred_gene_expr.txt
#and chr1_ALL_2_CAU_thrombomodulin_rankplt5_pheno_knn_pred_gene_expr.txt
ypred_frame_rf.to_csv("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"_rf_pred_gene_expr.txt", header=True, index=True, sep="\t")
ypred_frame_svr.to_csv("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"_svr_pred_gene_expr.txt", header=True, index=True, sep="\t")
ypred_frame_knn.to_csv("/home/pokoro/data/mesa_models/mesa_pheno/thrombotic/pred_expr/chunk/chr"+str(chrom)+"_chunk"+chunk+"_"+tr_pop+"_2_"+pop+"_knn_pred_gene_expr.txt", header=True, index=True, sep="\t")