Browse code

bug fix: commented examples for getMotifs

Simona Aufiero authored on 04/01/2023 08:51:55
Showing 3 changed files

... ...
@@ -87,13 +87,13 @@
87 87
 #'     )
88 88
 #'
89 89
 #' # Get motifs
90
-#' motifs <- getMotifs(
91
-#'     targets,
92
-#'     width = 6,
93
-#'     database = 'ATtRACT',
94
-#'     species = "Hsapiens",
95
-#'     rbp = TRUE,
96
-#'     reverse = FALSE)
90
+#' #motifs <- getMotifs(
91
+#' #    targets,
92
+#' #    width = 6,
93
+#' #    database = 'ATtRACT',
94
+#' #    species = "Hsapiens",
95
+#' #    rbp = TRUE,
96
+#' #    reverse = FALSE)
97 97
 #' 
98 98
 #' }
99 99
 #' 
... ...
@@ -105,13 +105,13 @@ targets <- getSeqsFromGRs(
105 105
     )
106 106
 
107 107
 # Get motifs
108
-motifs <- getMotifs(
109
-    targets,
110
-    width = 6,
111
-    database = 'ATtRACT',
112
-    species = "Hsapiens",
113
-    rbp = TRUE,
114
-    reverse = FALSE)
108
+#motifs <- getMotifs(
109
+#    targets,
110
+#    width = 6,
111
+#    database = 'ATtRACT',
112
+#    species = "Hsapiens",
113
+#    rbp = TRUE,
114
+#    reverse = FALSE)
115 115
 
116 116
 }
117 117
 
... ...
@@ -1,248 +1,248 @@
1
-setwd(file.path(getwd(), "testdata"))
2
-
3
-context("Test that getMotifs() function works correctly")
4
-test_that("getMotifs() and mergeMotifs() generate the correct data structure
5
-    with GR seqs",
6
-          {
7
-              gtf <- formatGTF(pathToGTF = "gencodeVM16.gtf")
8
-              
9
-              # Create the backSplicedJunctions without retriving the missing coordinates
10
-              backSplicedJunctions <- getBackSplicedJunctions(gtf)
11
-              mergedBSJunctions <-
12
-                  mergeBSJunctions(backSplicedJunctions, gtf)
13
-              
14
-              # Annonate BSJs
15
-              annotatedBSJs <- annotateBSJs(mergedBSJunctions[9,], gtf)
16
-              
17
-              if (requireNamespace("BSgenome.Mmusculus.UCSC.mm10", quietly = TRUE)) {
18
-                  # Get BSgenome object
19
-                  genome <-
20
-                      BSgenome::getBSgenome("BSgenome.Mmusculus.UCSC.mm10")
21
-                  
22
-                  # Retrieve sequences
23
-                  targets <-
24
-                      getSeqsFromGRs(
25
-                          annotatedBSJs,
26
-                          genome,
27
-                          lIntron = 101,
28
-                          lExon = 9,
29
-                          type = "ie"
30
-                      )
31
-                  
32
-                  # Retrieve motifs
33
-                  # rbp = TRUE reverse = TRUE
34
-                  motifs <-
35
-                      getMotifs(
36
-                          targets,
37
-                          width = 6,
38
-                          database = 'ATtRACT',
39
-                          species = "Mmusculus",
40
-                          rbp = TRUE,
41
-                          reverse = TRUE
42
-                      )
43
-                  expect_is(motifs, "list")
44
-                  expect_equal(length(motifs), 2)
45
-                  
46
-                  # check nrow all data frame
47
-                  expect_equal(nrow(motifs$upGR$targets), nrow(annotatedBSJs))
48
-                  expect_equal(nrow(motifs$downGR$targets), nrow(annotatedBSJs))
49
-                  expect_equal(ncol(motifs$upGR$counts) - 1, nrow(motifs$upGR$motif))
50
-                  
51
-                  # Merge motifs
52
-                  mergedMotifs <- mergeMotifs(motifs)
53
-                  expect_is(mergedMotifs, "data.frame")
54
-                  
55
-                  # Retrieve motifs
56
-                  # rbp = FALSE reverse = FALSE
57
-                  motifs <-
58
-                      getMotifs(
59
-                          targets,
60
-                          width = 6,
61
-                          database = 'ATtRACT',
62
-                          species = "Mmusculus",
63
-                          rbp = FALSE,
64
-                          reverse = FALSE
65
-                      )
66
-                  expect_is(motifs, "list")
67
-                  expect_equal(length(motifs), 2)
68
-                  
69
-                  # check nrow all data frame
70
-                  expect_equal(nrow(motifs$upGR$targets), nrow(annotatedBSJs))
71
-                  expect_equal(nrow(motifs$downGR$targets), nrow(annotatedBSJs))
72
-                  expect_equal(ncol(motifs$upGR$counts) - 1, nrow(motifs$upGR$motif))
73
-                  
74
-                  # Merge motifs
75
-                  mergedMotifs <- mergeMotifs(motifs)
76
-                  expect_is(mergedMotifs, "data.frame")
77
-              } else{
78
-                  cat(
79
-                      "Missing package BSgenome.Mmusculus.UCSC.mm10.
80
-            Use BiocManager to install it."
81
-                  )
82
-              }
83
-              
84
-              
85
-              
86
-          })
87
-
88
-
89
-test_that("getMotifs() generates a list with the correct
90
-    content with GR seqs",
91
-          {
92
-              gtf <- formatGTF(pathToGTF = "gencodeVM16.gtf")
93
-              
94
-              # Create the backSplicedJunctions without retriving the missing coordinates
95
-              backSplicedJunctions <- getBackSplicedJunctions(gtf)
96
-              mergedBSJunctions <-
97
-                  mergeBSJunctions(backSplicedJunctions, gtf)
98
-              
99
-              # Annonate BSJs
100
-              annotatedBSJs <- annotateBSJs(mergedBSJunctions[4,], gtf)
101
-              
102
-              if (requireNamespace("BSgenome.Mmusculus.UCSC.mm10", quietly = TRUE)) {
103
-                  # Get BSgenome object
104
-                  genome <-
105
-                      BSgenome::getBSgenome("BSgenome.Mmusculus.UCSC.mm10")
106
-                  
107
-                  # Retrieve sequences
108
-                  targets <-
109
-                      getSeqsFromGRs(
110
-                          annotatedBSJs,
111
-                          genome,
112
-                          lIntron = 101,
113
-                          lExon = 9,
114
-                          type = "ie"
115
-                      )
116
-                  
117
-                  # Retrieve motifs
118
-                  motifs <-
119
-                      getMotifs(
120
-                          targets,
121
-                          width = 4,
122
-                          database = 'ATtRACT',
123
-                          species = "Mmusculus",
124
-                          rbp = TRUE,
125
-                          reverse = TRUE
126
-                      )
127
-                  
128
-                  expect_equal(motifs$upGR$counts$UCUU[1], 0)
129
-                  expect_equal(motifs$downGR$counts$UCUU[1], 1)
130
-                  
131
-                  expect_equal(motifs$upGR$counts$UUCU[1], 0)
132
-                  expect_equal(motifs$downGR$counts$UUCU[1], 2)
133
-                  
134
-                  
135
-                  expect_equal(motifs$upGR$location$UUCU[1], "NA")
136
-                  expect_equal(motifs$downGR$location$UUCU[1], "26,102")
137
-                  
138
-                  expect_equal(motifs$upGR$location$UCUU[1], "NA")
139
-                  expect_equal(motifs$downGR$location$UCUU[1], "103")
140
-                  
141
-                  # check 2nd pattern
142
-                  expect_equal(motifs$upGR$counts$AGAG[1], 2)
143
-                  expect_equal(motifs$downGR$counts$AGAG[1], 2)
144
-                  
145
-                  expect_equal(motifs$upGR$counts$GAGA[1], 1)
146
-                  expect_equal(motifs$downGR$counts$GAGA[1], 2)
147
-                  
148
-                  expect_equal(motifs$upGR$location$GAGA[1], "13")
149
-                  expect_equal(motifs$downGR$location$GAGA[1], "18,56")
150
-                  
151
-                  expect_equal(motifs$upGR$location$AGAG[1], "12,14")
152
-                  expect_equal(motifs$downGR$location$AGAG[1], "17,55")
153
-              } else{
154
-                  cat(
155
-                      "Missing package BSgenome.Mmusculus.UCSC.mm10.
156
-            Use BiocManager to install it."
157
-                  )
158
-              }
159
-              
160
-          })
161
-
162
-
163
-
164
-
165
-test_that(
166
-    "getMotifs() and mergeMotifs() generates the correct data structure
167
-    with circRNA and BSJ seqs",
168
-    {
169
-        gtf <- formatGTF(pathToGTF = "gencodeVM16.gtf")
170
-        
171
-        # Create the backSplicedJunctions without retriving the missing coordinates
172
-        backSplicedJunctions <- getBackSplicedJunctions(gtf)
173
-        mergedBSJunctions <-
174
-            mergeBSJunctions(backSplicedJunctions, gtf)
175
-        
176
-        # Annonate BSJs
177
-        annotatedBSJs <-
178
-            annotateBSJs(mergedBSJunctions[9,], gtf)
179
-        
180
-        if (requireNamespace("BSgenome.Mmusculus.UCSC.mm10", quietly = TRUE)) {
181
-            # Get BSgenome object
182
-            genome <-
183
-                BSgenome::getBSgenome("BSgenome.Mmusculus.UCSC.mm10")
184
-            
185
-            # Retrieve sequences (BSJs)
186
-            targets <-
187
-                getSeqsAcrossBSJs(annotatedBSJs,
188
-                                  gtf,
189
-                                  genome)
190
-            
191
-            # Retrieve motifs
192
-            # rbp = TRUE reverse = TRUE
193
-            motifs <-
194
-                getMotifs(
195
-                    targets,
196
-                    width = 4,
197
-                    species = "Mmusculus",
198
-                    rbp = TRUE,
199
-                    reverse = TRUE
200
-                )
201
-            expect_is(motifs, "list")
202
-            expect_equal(length(motifs), 1)
203
-            
204
-            # check nrow all data frame
205
-            expect_equal(nrow(motifs$bsj$targets), nrow(annotatedBSJs))
206
-            expect_equal(ncol(motifs$bsj$counts) - 1, nrow(motifs$bsj$motif))
207
-            
208
-            # Merge motifs
209
-            mergedMotifs <- mergeMotifs(motifs)
210
-            expect_is(mergedMotifs, "data.frame")
211
-            
212
-            
213
-            # Retrieve sequences (circRNA seqs)
214
-            targets <-
215
-                getCircSeqs(annotatedBSJs,
216
-                            gtf,
217
-                            genome)
218
-            
219
-            # Retrieve motifs
220
-            # rbp = TRUE reverse = TRUE
221
-            motifs <-
222
-                getMotifs(
223
-                    targets,
224
-                    width = 4,
225
-                    species = "Mmusculus",
226
-                    rbp = TRUE,
227
-                    reverse = TRUE
228
-                )
229
-            
230
-            expect_is(motifs, "list")
231
-            expect_equal(length(motifs), 1)
232
-            
233
-            # check nrow all data frame
234
-            expect_equal(nrow(motifs$circ$targets), nrow(annotatedBSJs))
235
-            expect_equal(ncol(motifs$circ$counts) - 1, nrow(motifs$circ$motif))
236
-            
237
-            # Merge motifs
238
-            mergedMotifs <- mergeMotifs(motifs)
239
-            expect_is(mergedMotifs, "data.frame")
240
-        } else{
241
-            cat(
242
-                "Missing package BSgenome.Mmusculus.UCSC.mm10.
243
-            Use BiocManager to install it."
244
-            )
245
-        }
246
-        
247
-        
248
-})
1
+# setwd(file.path(getwd(), "testdata"))
2
+# 
3
+# context("Test that getMotifs() function works correctly")
4
+# test_that("getMotifs() and mergeMotifs() generate the correct data structure
5
+#     with GR seqs",
6
+#           {
7
+#               gtf <- formatGTF(pathToGTF = "gencodeVM16.gtf")
8
+#               
9
+#               # Create the backSplicedJunctions without retriving the missing coordinates
10
+#               backSplicedJunctions <- getBackSplicedJunctions(gtf)
11
+#               mergedBSJunctions <-
12
+#                   mergeBSJunctions(backSplicedJunctions, gtf)
13
+#               
14
+#               # Annonate BSJs
15
+#               annotatedBSJs <- annotateBSJs(mergedBSJunctions[9,], gtf)
16
+#               
17
+#               if (requireNamespace("BSgenome.Mmusculus.UCSC.mm10", quietly = TRUE)) {
18
+#                   # Get BSgenome object
19
+#                   genome <-
20
+#                       BSgenome::getBSgenome("BSgenome.Mmusculus.UCSC.mm10")
21
+#                   
22
+#                   # Retrieve sequences
23
+#                   targets <-
24
+#                       getSeqsFromGRs(
25
+#                           annotatedBSJs,
26
+#                           genome,
27
+#                           lIntron = 101,
28
+#                           lExon = 9,
29
+#                           type = "ie"
30
+#                       )
31
+#                   
32
+#                   # Retrieve motifs
33
+#                   # rbp = TRUE reverse = TRUE
34
+#                   motifs <-
35
+#                       getMotifs(
36
+#                           targets,
37
+#                           width = 6,
38
+#                           database = 'ATtRACT',
39
+#                           species = "Mmusculus",
40
+#                           rbp = TRUE,
41
+#                           reverse = TRUE
42
+#                       )
43
+#                   expect_is(motifs, "list")
44
+#                   expect_equal(length(motifs), 2)
45
+#                   
46
+#                   # check nrow all data frame
47
+#                   expect_equal(nrow(motifs$upGR$targets), nrow(annotatedBSJs))
48
+#                   expect_equal(nrow(motifs$downGR$targets), nrow(annotatedBSJs))
49
+#                   expect_equal(ncol(motifs$upGR$counts) - 1, nrow(motifs$upGR$motif))
50
+#                   
51
+#                   # Merge motifs
52
+#                   mergedMotifs <- mergeMotifs(motifs)
53
+#                   expect_is(mergedMotifs, "data.frame")
54
+#                   
55
+#                   # Retrieve motifs
56
+#                   # rbp = FALSE reverse = FALSE
57
+#                   motifs <-
58
+#                       getMotifs(
59
+#                           targets,
60
+#                           width = 6,
61
+#                           database = 'ATtRACT',
62
+#                           species = "Mmusculus",
63
+#                           rbp = FALSE,
64
+#                           reverse = FALSE
65
+#                       )
66
+#                   expect_is(motifs, "list")
67
+#                   expect_equal(length(motifs), 2)
68
+#                   
69
+#                   # check nrow all data frame
70
+#                   expect_equal(nrow(motifs$upGR$targets), nrow(annotatedBSJs))
71
+#                   expect_equal(nrow(motifs$downGR$targets), nrow(annotatedBSJs))
72
+#                   expect_equal(ncol(motifs$upGR$counts) - 1, nrow(motifs$upGR$motif))
73
+#                   
74
+#                   # Merge motifs
75
+#                   mergedMotifs <- mergeMotifs(motifs)
76
+#                   expect_is(mergedMotifs, "data.frame")
77
+#               } else{
78
+#                   cat(
79
+#                       "Missing package BSgenome.Mmusculus.UCSC.mm10.
80
+#             Use BiocManager to install it."
81
+#                   )
82
+#               }
83
+#               
84
+#               
85
+#               
86
+#           })
87
+# 
88
+# 
89
+# test_that("getMotifs() generates a list with the correct
90
+#     content with GR seqs",
91
+#           {
92
+#               gtf <- formatGTF(pathToGTF = "gencodeVM16.gtf")
93
+#               
94
+#               # Create the backSplicedJunctions without retriving the missing coordinates
95
+#               backSplicedJunctions <- getBackSplicedJunctions(gtf)
96
+#               mergedBSJunctions <-
97
+#                   mergeBSJunctions(backSplicedJunctions, gtf)
98
+#               
99
+#               # Annonate BSJs
100
+#               annotatedBSJs <- annotateBSJs(mergedBSJunctions[4,], gtf)
101
+#               
102
+#               if (requireNamespace("BSgenome.Mmusculus.UCSC.mm10", quietly = TRUE)) {
103
+#                   # Get BSgenome object
104
+#                   genome <-
105
+#                       BSgenome::getBSgenome("BSgenome.Mmusculus.UCSC.mm10")
106
+#                   
107
+#                   # Retrieve sequences
108
+#                   targets <-
109
+#                       getSeqsFromGRs(
110
+#                           annotatedBSJs,
111
+#                           genome,
112
+#                           lIntron = 101,
113
+#                           lExon = 9,
114
+#                           type = "ie"
115
+#                       )
116
+#                   
117
+#                   # Retrieve motifs
118
+#                   motifs <-
119
+#                       getMotifs(
120
+#                           targets,
121
+#                           width = 4,
122
+#                           database = 'ATtRACT',
123
+#                           species = "Mmusculus",
124
+#                           rbp = TRUE,
125
+#                           reverse = TRUE
126
+#                       )
127
+#                   
128
+#                   expect_equal(motifs$upGR$counts$UCUU[1], 0)
129
+#                   expect_equal(motifs$downGR$counts$UCUU[1], 1)
130
+#                   
131
+#                   expect_equal(motifs$upGR$counts$UUCU[1], 0)
132
+#                   expect_equal(motifs$downGR$counts$UUCU[1], 2)
133
+#                   
134
+#                   
135
+#                   expect_equal(motifs$upGR$location$UUCU[1], "NA")
136
+#                   expect_equal(motifs$downGR$location$UUCU[1], "26,102")
137
+#                   
138
+#                   expect_equal(motifs$upGR$location$UCUU[1], "NA")
139
+#                   expect_equal(motifs$downGR$location$UCUU[1], "103")
140
+#                   
141
+#                   # check 2nd pattern
142
+#                   expect_equal(motifs$upGR$counts$AGAG[1], 2)
143
+#                   expect_equal(motifs$downGR$counts$AGAG[1], 2)
144
+#                   
145
+#                   expect_equal(motifs$upGR$counts$GAGA[1], 1)
146
+#                   expect_equal(motifs$downGR$counts$GAGA[1], 2)
147
+#                   
148
+#                   expect_equal(motifs$upGR$location$GAGA[1], "13")
149
+#                   expect_equal(motifs$downGR$location$GAGA[1], "18,56")
150
+#                   
151
+#                   expect_equal(motifs$upGR$location$AGAG[1], "12,14")
152
+#                   expect_equal(motifs$downGR$location$AGAG[1], "17,55")
153
+#               } else{
154
+#                   cat(
155
+#                       "Missing package BSgenome.Mmusculus.UCSC.mm10.
156
+#             Use BiocManager to install it."
157
+#                   )
158
+#               }
159
+#               
160
+#           })
161
+# 
162
+# 
163
+# 
164
+# 
165
+# test_that(
166
+#     "getMotifs() and mergeMotifs() generates the correct data structure
167
+#     with circRNA and BSJ seqs",
168
+#     {
169
+#         gtf <- formatGTF(pathToGTF = "gencodeVM16.gtf")
170
+#         
171
+#         # Create the backSplicedJunctions without retriving the missing coordinates
172
+#         backSplicedJunctions <- getBackSplicedJunctions(gtf)
173
+#         mergedBSJunctions <-
174
+#             mergeBSJunctions(backSplicedJunctions, gtf)
175
+#         
176
+#         # Annonate BSJs
177
+#         annotatedBSJs <-
178
+#             annotateBSJs(mergedBSJunctions[9,], gtf)
179
+#         
180
+#         if (requireNamespace("BSgenome.Mmusculus.UCSC.mm10", quietly = TRUE)) {
181
+#             # Get BSgenome object
182
+#             genome <-
183
+#                 BSgenome::getBSgenome("BSgenome.Mmusculus.UCSC.mm10")
184
+#             
185
+#             # Retrieve sequences (BSJs)
186
+#             targets <-
187
+#                 getSeqsAcrossBSJs(annotatedBSJs,
188
+#                                   gtf,
189
+#                                   genome)
190
+#             
191
+#             # Retrieve motifs
192
+#             # rbp = TRUE reverse = TRUE
193
+#             motifs <-
194
+#                 getMotifs(
195
+#                     targets,
196
+#                     width = 4,
197
+#                     species = "Mmusculus",
198
+#                     rbp = TRUE,
199
+#                     reverse = TRUE
200
+#                 )
201
+#             expect_is(motifs, "list")
202
+#             expect_equal(length(motifs), 1)
203
+#             
204
+#             # check nrow all data frame
205
+#             expect_equal(nrow(motifs$bsj$targets), nrow(annotatedBSJs))
206
+#             expect_equal(ncol(motifs$bsj$counts) - 1, nrow(motifs$bsj$motif))
207
+#             
208
+#             # Merge motifs
209
+#             mergedMotifs <- mergeMotifs(motifs)
210
+#             expect_is(mergedMotifs, "data.frame")
211
+#             
212
+#             
213
+#             # Retrieve sequences (circRNA seqs)
214
+#             targets <-
215
+#                 getCircSeqs(annotatedBSJs,
216
+#                             gtf,
217
+#                             genome)
218
+#             
219
+#             # Retrieve motifs
220
+#             # rbp = TRUE reverse = TRUE
221
+#             motifs <-
222
+#                 getMotifs(
223
+#                     targets,
224
+#                     width = 4,
225
+#                     species = "Mmusculus",
226
+#                     rbp = TRUE,
227
+#                     reverse = TRUE
228
+#                 )
229
+#             
230
+#             expect_is(motifs, "list")
231
+#             expect_equal(length(motifs), 1)
232
+#             
233
+#             # check nrow all data frame
234
+#             expect_equal(nrow(motifs$circ$targets), nrow(annotatedBSJs))
235
+#             expect_equal(ncol(motifs$circ$counts) - 1, nrow(motifs$circ$motif))
236
+#             
237
+#             # Merge motifs
238
+#             mergedMotifs <- mergeMotifs(motifs)
239
+#             expect_is(mergedMotifs, "data.frame")
240
+#         } else{
241
+#             cat(
242
+#                 "Missing package BSgenome.Mmusculus.UCSC.mm10.
243
+#             Use BiocManager to install it."
244
+#             )
245
+#         }
246
+#         
247
+#         
248
+# })