|
4
|
1 #I should make an SOP for this. Problems we encountered: no x in the xY motif, and the kilodemon
|
|
|
2 #the output files have both Y and xY, they shouldn't why is that happening? make it not happen
|
|
|
3 #make sure that accession numbers stay locked to each motif, somehow
|
|
|
4 #output should look just like the KALIP input
|
|
|
5 #S2 is negatives
|
|
|
6 #ff you want ONLY FULL MOTIFS, put "YES" here, please use all caps
|
|
|
7 FullMotifsOnly_questionmark<-"NO"
|
|
|
8 #If you want ONLY TRUNCATED MOTIFS, put "YES" here, please use all caps
|
|
|
9 TruncatedMotifsOnly_questionmark<-"NO"
|
|
|
10 #if you want to find the overlap, put a "YES" here (all caps), if you want to find the non-overlap, put "NO" (all caps)
|
|
|
11 Are_You_Looking_For_Commonality<-"NO"
|
|
|
12
|
|
|
13
|
|
|
14 #put the names of your input files here
|
|
|
15 FirstSubstrateSet<- read.csv("Substrates 1A TiO2 and FeNTA no duplicates.csv", stringsAsFactors=FALSE)
|
|
|
16 Firstsubbackfreq<- read.csv("SBF 1A TiO2 and FeNTA no duplicates.csv", header=FALSE, stringsAsFactors=FALSE)
|
|
|
17
|
|
|
18 SecondSubstrateSet<- read.csv("Substrates 1B TiO2 and FeNTA no duplicates.csv", stringsAsFactors=FALSE)
|
|
|
19 Secondsubbackfreq<- read.csv("SBF 1B TiO2 and FeNTA no duplicates.csv", header=FALSE, stringsAsFactors=FALSE)
|
|
|
20
|
|
|
21 # ThirdSubstrateSet<- read.csv("Galaxy69-BTK_PLUS_R3_Substrates.csv", stringsAsFactors=FALSE)
|
|
|
22 # Thirdsubbackfreq<- read.csv("Galaxy70-BTK_PLUS_R3_SubstrateBackgroundFrequency.csv", header=FALSE, stringsAsFactors=FALSE)
|
|
|
23
|
|
|
24 #then put the names of your output files here
|
|
|
25 # Shared_motifs_table<-"180719_GALAXY-BTK-plus-rep-OVLP-7to7-substrates.csv"
|
|
|
26 # Shared_subbackfreq_table<-"180719_GALAXY-BTK-plus-rep-OVLP-7to7-SubBackFreq.csv"
|
|
|
27
|
|
|
28 # Shared_motifs_table<-"Shared motifs 7-27-17.csv"
|
|
|
29 # Shared_subbackfreq_table<-"SubstrateBackgrounFrequency-for-shared-motifs 4 7-27-17.csv"
|
|
|
30
|
|
|
31 First_unshared_motifs_table<-"pka_minus_wo-plus_wo_vector_subs.csv"
|
|
|
32 First_unshared_subbackfreq<-"pka_minus_wo-plus_wo_vector_sbf.csv"
|
|
|
33
|
|
|
34 Second_unshared_motifs_table<-"pka_plus_wo-minus_wo_vector_subs.csv"
|
|
|
35 Second_unshared_subbackfreq<-"pka_plus_wo-minus_wo_vector_sbf.csv"
|
|
|
36
|
|
|
37 # Third_unshared_motifs_table<-"R3 subs.csv"
|
|
|
38 # Third_unshared_subbackfreq<-"R3 SBF.csv"
|
|
|
39
|
|
|
40 #final note, this code is going to be unworkable if you want to make a Venn diagram of more than 3 circles. I think I'll poke around
|
|
|
41 #other languages to see if any of them can do it.
|
|
|
42 ####################################################################################################################################
|
|
|
43
|
|
|
44
|
|
|
45
|
|
|
46
|
|
|
47
|
|
|
48 # grepl(pattern = "S", x=asdf, ignore.case = TRUE)
|
|
|
49
|
|
|
50 FirstCentralLetters<-FirstSubstrateSet[,11]
|
|
|
51 SecondCentralLetters<-SecondSubstrateSet[,11]
|
|
|
52
|
|
|
53 FirstEsses<-sapply(FirstCentralLetters, grepl, pattern="S", ignore.case=TRUE)
|
|
|
54 FirstTees<-sapply(FirstCentralLetters, grepl, pattern="T", ignore.case=TRUE)
|
|
|
55 FirstWys<-sapply(FirstCentralLetters, grepl, pattern="Y", ignore.case=TRUE)
|
|
|
56
|
|
|
57 SecondEsses<-sapply(SecondCentralLetters, grepl, pattern="S", ignore.case=TRUE)
|
|
|
58 SecondTees<-sapply(SecondCentralLetters, grepl, pattern="T", ignore.case=TRUE)
|
|
|
59 SecondWys<-sapply(SecondCentralLetters, grepl, pattern="Y", ignore.case=TRUE)
|
|
|
60
|
|
|
61 FirstCentralLetters<-replace(FirstCentralLetters,FirstEsses,"xS")
|
|
|
62 FirstCentralLetters<-replace(FirstCentralLetters,FirstTees,"xT")
|
|
|
63 FirstCentralLetters<-replace(FirstCentralLetters,FirstWys,"xY")
|
|
|
64
|
|
|
65 SecondCentralLetters<-replace(SecondCentralLetters,SecondEsses,"xS")
|
|
|
66 SecondCentralLetters<-replace(SecondCentralLetters,SecondTees,"xT")
|
|
|
67 SecondCentralLetters<-replace(SecondCentralLetters,SecondWys,"xY")
|
|
|
68
|
|
|
69 FirstCentralLetters->FirstSubstrateSet[,11]
|
|
|
70 SecondCentralLetters->SecondSubstrateSet[,11]
|
|
|
71
|
|
|
72 ####################################################################################################################################
|
|
|
73 ####################################################################################################################################
|
|
|
74 # better version of this code written in C: what happens when two kinases share a motif, but they found that motif in two
|
|
|
75 # separate proteins thus two separate accession numbers?
|
|
|
76 # It should actually output the shared motif and BOTH accession numbers. Right now it does not, it only maps out the second
|
|
|
77 # accession number. So that needs to be fixed BUT you need to keep the commonality between a motif and its accession number
|
|
|
78 ####################################################################################################################################
|
|
|
79 ####################################################################################################################################
|
|
|
80 ####################################################################################################################################
|
|
|
81 ####################################################################################################################################
|
|
|
82
|
|
|
83 #Create the motif sets, deciding wether or not you're looking for truncated or full here
|
|
|
84 #full only
|
|
|
85
|
|
|
86
|
|
|
87 if (Are_You_Looking_For_Commonality=="NO"){
|
|
|
88
|
|
|
89 ###############################################
|
|
|
90 #ALL motifs, full and truncated
|
|
|
91
|
|
|
92 if (FullMotifsOnly_questionmark!="YES"&&TruncatedMotifsOnly_questionmark!="YES"){
|
|
|
93 # FTLwtmotifs=matrix(,nrow = nrow(FirstSubstrateSet),ncol=1)
|
|
|
94 # FTLwtAccessionNumbers=matrix(,nrow = nrow(FirstSubstrateSet),ncol=1)
|
|
|
95
|
|
|
96 #so what this does is
|
|
|
97 #find the central phospho position and add spaces to either side of it
|
|
|
98 #create a pasted together motif which will be used for difference finding
|
|
|
99 #keep motifs and accession numbers in the same index position of two vectors
|
|
|
100
|
|
|
101 #so instead just take the FLTwt[i,4:18], and replace "" with "O"
|
|
|
102 #then perform paste function
|
|
|
103 #then FLTwt with "O" can be the motif that will get printed to the CSV (after re-removing the "O") and the pasted motif is in its own vector that follows the A #s
|
|
|
104
|
|
|
105 FTLwtmotifs<-FirstSubstrateSet[,4:18]
|
|
|
106 # FTLwtmotifs<-sapply(FTLwtmotifs, gsub, pattern=" ", replacement="O",ignore.case=TRUE)
|
|
|
107 FTLwtmotifs<-replace(FTLwtmotifs, FTLwtmotifs=="", "O")
|
|
|
108 FTLwtmotifs<-apply(FTLwtmotifs, 1, paste, sep="", collapse="")
|
|
|
109 FTLwtAccessionNumbers<-FirstSubstrateSet[,3]
|
|
|
110
|
|
|
111 # D835Ymotifs=matrix(,nrow = nrow(SecondSubstrateSet),ncol=1)
|
|
|
112 # D835YAccessionNumbers<-matrix(,nrow = nrow(SecondSubstrateSet),ncol = 1)
|
|
|
113
|
|
|
114 D835Ymotifs<-SecondSubstrateSet[,4:18]
|
|
|
115 # D835Ymotifs<-sapply(D835Ymotifs, gsub, patter="", replacement="O",ignore.case=TRUE)
|
|
|
116 D835Ymotifs<-replace(D835Ymotifs, D835Ymotifs=="", "O")
|
|
|
117 D835Ymotifs<-replace(D835Ymotifs, D835Ymotifs=="", "O")
|
|
|
118 D835Ymotifs<-apply(D835Ymotifs, 1, paste, sep="", collapse="")
|
|
|
119 D835YAccessionNumbers<-SecondSubstrateSet[,3]
|
|
|
120
|
|
|
121 names(FTLwtmotifs)<-FTLwtAccessionNumbers
|
|
|
122 names(D835Ymotifs)<-D835YAccessionNumbers
|
|
|
123
|
|
|
124 }
|
|
|
125
|
|
|
126
|
|
|
127 FTLwtmotifsFINAL<-FTLwtmotifs[!FTLwtmotifs %in% D835Ymotifs]
|
|
|
128 FTLwtmotifsFINAL<-FTLwtmotifsFINAL[!duplicated(FTLwtmotifsFINAL)]
|
|
|
129
|
|
|
130 D835YmotifsFINAL<-D835Ymotifs[!D835Ymotifs %in% FTLwtmotifs]
|
|
|
131 D835YmotifsFINAL<-D835YmotifsFINAL[!duplicated(D835YmotifsFINAL)]
|
|
|
132
|
|
|
133
|
|
|
134
|
|
|
135 # columnalheader<-c(rep(NA,36))
|
|
|
136 # FTLFinalMatrix<-matrix(data =columnalheader,nrow = 1)
|
|
|
137 #
|
|
|
138 #how to make this an apply
|
|
|
139 #actual first turn names(motifsFINAL) into a vector, then gsub to replace all "|" pipe characters with "*" star characters
|
|
|
140 #first take the accession numbers as Firstsubbackfreq[1,], then gsub to replace all "|" pipe characters with "*" star characters
|
|
|
141 #then use AccessionNumbers as a vector and use
|
|
|
142 #matches <- unique (grep(paste(toMatch,collapse="|"), names(motifsFINAL), value=TRUE))
|
|
|
143
|
|
|
144 FLTwtNames<-names(FTLwtmotifsFINAL)
|
|
|
145 FLTwtNames<-FLTwtNames[!duplicated(FLTwtNames)]
|
|
|
146 #the problem is the names includes a lot of things that are multi names separated by ;
|
|
|
147 #I need to do soem paper writing and poster writing
|
|
|
148
|
|
|
149 FLTwtNames<-gsub(pattern = '\\|', replacement = "*", x=FLTwtNames)
|
|
|
150 FLTwtNames<-paste(FLTwtNames,collapse = "|")
|
|
|
151
|
|
|
152 AccNums<-as.character(Firstsubbackfreq[1,])
|
|
|
153 AccNums<-gsub(pattern = "\\|", replacement = "*", x=AccNums)
|
|
|
154
|
|
|
155
|
|
|
156 firstthree123<-gsub(pattern = "\\|", replacement = "", x=FLTwtNames)
|
|
|
157 lastthree123<-gsub(pattern = "\\|", replacement = "", x=AccNums)
|
|
|
158
|
|
|
159 pasty<-paste(lastthree123,collapse = "|")
|
|
|
160
|
|
|
161 my_matches<-unique(grep(pattern = FLTwtNames, x=AccNums, value = TRUE))
|
|
|
162 # matches <- unique (grep(paste(FLTwtNames,collapse="|"), AccNums, value=TRUE))
|
|
|
163 My_matches<- AccNums %in% FLTwtNames #use apply grep function
|
|
|
164 My_matches<-grep(pattern = pasty, x=firstthree123, ignore.case = TRUE)
|
|
|
165 sensation<-firstthree123[My_matches]
|
|
|
166 what<-firstthree123[!firstthree123 %in% sensation]
|
|
|
167
|
|
|
168 #my matches are which columns in the SBF file still have a motif with them
|
|
|
169 FirstFINALSBF<-Firstsubbackfreq[My_matches]
|
|
|
170 FTLFinalMatrix<-FirstFINALSBF
|
|
|
171 duped<-FTLFinalMatrix[1,]
|
|
|
172 duped<-duped[!duplicated(duped)]
|
|
|
173 names(duped)<-NULL
|
|
|
174 duped<-unlist(duped)
|
|
|
175 unduped<-FTLFinalMatrix[,!duplicated(duped)]
|
|
|
176 FTLFinalMatrix<-unduped
|
|
|
177 # FTLFinalMatrix<-FTLFinalMatrix[2:nrow(FTLFinalMatrix),]
|
|
|
178
|
|
|
179 # columnalheader<-c(rep(NA,36))
|
|
|
180 # D835YFinalMatrix<-matrix(data =columnalheader,nrow = 1)
|
|
|
181
|
|
|
182 D835names<-names(D835YmotifsFINAL)
|
|
|
183 D835names<-D835names[!duplicated(D835names)]
|
|
|
184 AccNums<-as.character(Secondsubbackfreq[1,])
|
|
|
185 My_matches<-AccNums %in% D835names
|
|
|
186 SecondFINALSBF<-Secondsubbackfreq[My_matches]
|
|
|
187 D835YFinalMatrix<-SecondFINALSBF
|
|
|
188 duped<-D835YFinalMatrix[1,]
|
|
|
189 duped<-duped[!duplicated(duped)]
|
|
|
190 names(duped)<-NULL
|
|
|
191 duped<-unlist(duped)
|
|
|
192 unduped<-D835YFinalMatrix[,!duplicated(duped)]
|
|
|
193 D835YFinalMatrix<-unduped
|
|
|
194 #there's a problem, accession numbers are duplicated
|
|
|
195
|
|
|
196 # FTLoutputmatrix<-matrix(data=c(FTLwtmotifsFINAL,names(FTLwtmotifsFINAL)),ncol = 2)
|
|
|
197
|
|
|
198 FTLwtmotifsFINAL<-gsub(pattern = "O", replacement = " ", x=FTLwtmotifsFINAL)
|
|
|
199 FTLwtmotifsFINAL<-gsub(pattern = "x", replacement = "", x=FTLwtmotifsFINAL)
|
|
|
200 FTLoutputmatrix<-matrix(data = c(FTLwtmotifsFINAL,names(FTLwtmotifsFINAL)), ncol=2)
|
|
|
201
|
|
|
202 # FTLwtmotifsFINAL<-gsub(pattern = "O", replacement = " ", x=FTLwtmotifsFINAL)
|
|
|
203 # FTLwtmotifsFINAL<-gsub(pattern = "x", replacement = "", x=FTLwtmotifsFINAL)
|
|
|
204 # FTLoutputmatrix<-matrix(data = c(FTLwtmotifsFINAL,names(FTLwtmotifsFINAL)), ncol=2)
|
|
|
205
|
|
|
206 FLTheader<-c("Substrate","Species","Reference","-7","-6","-5","-4","-3","-2","-1","0","1","2","3","4","5","6","7","Phosphite")
|
|
|
207 # FLTheader<-unlist(FLTheader)
|
|
|
208 lefthandFLT<-matrix(data = rep(NA,times=2*nrow(FTLoutputmatrix)),nrow=nrow(FTLoutputmatrix))
|
|
|
209 righthandFLT<-matrix(data = rep(NA,times=1*nrow(FTLoutputmatrix)),nrow=nrow(FTLoutputmatrix))
|
|
|
210 FLTaccessionset<-FTLoutputmatrix[,2]
|
|
|
211 FTLmeat<-sapply(FTLoutputmatrix[,1], strsplit, "")
|
|
|
212 FTLmeat<-sapply(FTLmeat, unlist)
|
|
|
213 colnames(FTLmeat)<-NULL
|
|
|
214 FTLmeat<-t(FTLmeat)
|
|
|
215
|
|
|
216 FTLoutputmatrix2<-cbind(lefthandFLT,FLTaccessionset,FTLmeat,righthandFLT)
|
|
|
217 colnames(FTLoutputmatrix2)<-NULL
|
|
|
218 rownames(FTLoutputmatrix2)<-NULL
|
|
|
219 colnames(FLTheader)<-NULL
|
|
|
220 rownames(FLTheader)<-NULL
|
|
|
221
|
|
|
222
|
|
|
223 FirstCentralLettersAGAIN<-FTLoutputmatrix2[,11]
|
|
|
224
|
|
|
225 FirstEsses<-sapply(FirstCentralLettersAGAIN, grepl, pattern="S", ignore.case=TRUE)
|
|
|
226 FirstTees<-sapply(FirstCentralLettersAGAIN, grepl, pattern="T", ignore.case=TRUE)
|
|
|
227 FirstWys<-sapply(FirstCentralLettersAGAIN, grepl, pattern="Y", ignore.case=TRUE)
|
|
|
228
|
|
|
229 FirstCentralLettersAGAIN<-replace(FirstCentralLettersAGAIN,FirstEsses,"xS")
|
|
|
230 FirstCentralLettersAGAIN<-replace(FirstCentralLettersAGAIN,FirstTees,"xT")
|
|
|
231 FirstCentralLettersAGAIN<-replace(FirstCentralLettersAGAIN,FirstWys,"xY")
|
|
|
232
|
|
|
233 FirstCentralLettersAGAIN->FTLoutputmatrix2[,11]
|
|
|
234
|
|
|
235 FTLoutputmatrix2<-rbind(FLTheader,FTLoutputmatrix2)
|
|
|
236
|
|
|
237 write.table(x=FTLoutputmatrix2,
|
|
|
238 file=First_unshared_motifs_table,
|
|
|
239 quote=FALSE, sep=",",
|
|
|
240 row.names=FALSE,col.names = FALSE, na="", append=TRUE)
|
|
|
241
|
|
|
242 columnalheader<-c("Accession Numbers",as.character(Firstsubbackfreq[1:35,1]))
|
|
|
243 columnalheader<-matrix(columnalheader,ncol = 1)
|
|
|
244 # write.table(x=columnalheader,
|
|
|
245 # file=First_unshared_subbackfreq,
|
|
|
246 # quote=FALSE, sep=",",
|
|
|
247 # row.names=FALSE,col.names = FALSE, na="", append=TRUE)
|
|
|
248
|
|
|
249 FTLFinalMatrix<-cbind(columnalheader,FTLFinalMatrix)
|
|
|
250
|
|
|
251 write.table(x=FTLFinalMatrix,
|
|
|
252 file=First_unshared_subbackfreq,
|
|
|
253 quote=FALSE, sep=",",
|
|
|
254 row.names=FALSE,col.names = FALSE, na="", append=TRUE)
|
|
|
255
|
|
|
256 ############################################################################################################
|
|
|
257
|
|
|
258 D835YmotifsFINAL<-gsub(pattern = "O", replacement = " ", x=D835YmotifsFINAL)
|
|
|
259 D835YmotifsFINAL<-gsub(pattern = "x", replacement = "", x=D835YmotifsFINAL)
|
|
|
260 D835Youtputmatrix<-matrix(data = c(D835YmotifsFINAL,names(D835YmotifsFINAL)), ncol=2)
|
|
|
261
|
|
|
262 # D835YmotifsFINAL<-gsub(pattern = "O", replacement = " ", x=D835YoutputmatrixL)
|
|
|
263 # D835YmotifsFINAL<-gsub(pattern = "x", replacement = "", x=D835Youtputmatrix)
|
|
|
264 # D835Youtputmatrix<-matrix(data = c(D835Youtputmatrix,names(D835Youtputmatrix)), ncol=2)
|
|
|
265 #
|
|
|
266
|
|
|
267 # D835Youtputmatrix<-matrix(data=c(D835YmotifsFINAL,names(D835YmotifsFINAL)),ncol = 2)
|
|
|
268
|
|
|
269 D835Yheader<-c("Substrate","Species","Reference","-7","-6","-5","-4","-3","-2","-1","0","1","2","3","4","5","6","7","Phosphite")
|
|
|
270 # D835Yheader<-unlist(D835Yheader)
|
|
|
271 lefthandD835<-matrix(data = rep(NA,times=2*nrow(D835Youtputmatrix)),nrow=nrow(D835Youtputmatrix))
|
|
|
272 righthandD835<-matrix(data = rep(NA,times=1*nrow(D835Youtputmatrix)),nrow=nrow(D835Youtputmatrix))
|
|
|
273 D835Yaset<-D835Youtputmatrix[,2]
|
|
|
274 D835meat<-sapply(D835Youtputmatrix[,1], strsplit, "")
|
|
|
275 D835meat<-sapply(D835meat, unlist)
|
|
|
276 colnames(D835meat)<-NULL
|
|
|
277 D835meat<-t(D835meat)
|
|
|
278
|
|
|
279 # FTLmeat<-sapply(FTLoutputmatrix[,1], strsplit, "")
|
|
|
280 # FTLmeat<-sapply(FTLmeat, unlist)
|
|
|
281 # colnames(FTLmeat)<-NULL
|
|
|
282 # FTLmeat<-t(FTLmeat)
|
|
|
283
|
|
|
284
|
|
|
285 D835Youtputmatrix2<-cbind(lefthandD835,D835Yaset,D835meat,righthandD835)
|
|
|
286 colnames(D835Youtputmatrix2)<-NULL
|
|
|
287 rownames(D835Youtputmatrix2)<-NULL
|
|
|
288 colnames(D835Yheader)<-NULL
|
|
|
289 rownames(D835Yheader)<-NULL
|
|
|
290
|
|
|
291
|
|
|
292 SecondCentralLettersAGAIN<-D835Youtputmatrix2[,11]
|
|
|
293
|
|
|
294 SecondEsses<-sapply(SecondCentralLettersAGAIN, grepl, pattern="S", ignore.case=TRUE)
|
|
|
295 SecondTees<-sapply(SecondCentralLettersAGAIN, grepl, pattern="T", ignore.case=TRUE)
|
|
|
296 SecondWys<-sapply(SecondCentralLettersAGAIN, grepl, pattern="Y", ignore.case=TRUE)
|
|
|
297
|
|
|
298 SecondCentralLettersAGAIN<-replace(SecondCentralLettersAGAIN,SecondEsses,"xS")
|
|
|
299 SecondCentralLettersAGAIN<-replace(SecondCentralLettersAGAIN,SecondTees,"xT")
|
|
|
300 SecondCentralLettersAGAIN<-replace(SecondCentralLettersAGAIN,SecondWys,"xY")
|
|
|
301
|
|
|
302 SecondCentralLettersAGAIN->D835Youtputmatrix2[,11]
|
|
|
303
|
|
|
304 D835Youtputmatrix2<-rbind(D835Yheader,D835Youtputmatrix2)
|
|
|
305
|
|
|
306 write.table(x=D835Youtputmatrix2,
|
|
|
307 file=Second_unshared_motifs_table,
|
|
|
308 quote=FALSE, sep=",",
|
|
|
309 row.names=FALSE,col.names = FALSE, na="", append=TRUE)
|
|
|
310
|
|
|
311 columnalheader<-c("Accession Numbers",as.character(Firstsubbackfreq[1:35,1]))
|
|
|
312 columnalheader<-matrix(columnalheader,ncol = 1)
|
|
|
313 # write.table(x=columnalheader,
|
|
|
314 # file=Second_unshared_subbackfreq,
|
|
|
315 # quote=FALSE, sep=",",
|
|
|
316 # row.names=FALSE,col.names = FALSE, na="", append=TRUE)
|
|
|
317
|
|
|
318 D835YFinalMatrix<-cbind(columnalheader,D835YFinalMatrix)
|
|
|
319
|
|
|
320 write.table(x=D835YFinalMatrix,
|
|
|
321 file=Second_unshared_subbackfreq,
|
|
|
322 quote=FALSE, sep=",",
|
|
|
323 row.names=FALSE,col.names = FALSE, na="", append=TRUE)
|
|
|
324 }
|
|
|
325
|
|
|
326
|
|
|
327
|
|
|
328 # for (i in 1:nrow(FirstSubstrateSet)){
|
|
|
329 # FTLwtletters<-FirstSubstrateSet[i,4:18]
|
|
|
330 # FTLwtletters<-FTLwtletters[FTLwtletters !="XXXXX"]
|
|
|
331 # FTLwtletters<-paste(FTLwtletters, sep="", collapse="")
|
|
|
332 # leftspaces<-c()
|
|
|
333 # rightspaces<-c()
|
|
|
334 #
|
|
|
335 # YYYmotif <- unlist(strsplit(FTLwtletters, split = ""))
|
|
|
336 # YYYposition <- match(x = "x", table = YYYmotif)
|
|
|
337 # #position itself tells me how much is to the left of that X by what it's number is. x at position 4 tells me that there are
|
|
|
338 # #just 3 letters to the left of x
|
|
|
339 #
|
|
|
340 # YYYLettersToTheLeft <- YYYposition - 1
|
|
|
341 # #how many letters to the right SHOULD just be length(motif)-position-1 if it's 5 long and x is at 3 then Y is at 4 and there is
|
|
|
342 # #just 1 spot to the right of Y so LettersToTheRight<-1 because 5-3-1=1
|
|
|
343 # YYYLettersToTheRight <- length(YYYmotif) - YYYposition - 1
|
|
|
344 # #then sanity check, we're currently looking only at +/-4, but this spot allows for up to +/- 7 as well, just depends on what the
|
|
|
345 # #variable the user puts in is
|
|
|
346 #
|
|
|
347 #
|
|
|
348 # if (YYYLettersToTheLeft < 7 | YYYLettersToTheRight < 7) {
|
|
|
349 # leftspaces<-rep(" ",times=(7-YYYLettersToTheLeft))
|
|
|
350 # rightspaces<-rep(" ",times=7-(YYYLettersToTheRight))
|
|
|
351 # #add blank spaces if the motif has less than 4 letters to the left/right
|
|
|
352 # motif<-c(leftspaces,YYYmotif,rightspaces)
|
|
|
353 # #save that motif, which is the Y and +/- 4 amino acids, including truncation
|
|
|
354 # motif<-motif[!motif %in% "x"]
|
|
|
355 # motif<-paste(motif, sep="", collapse="")
|
|
|
356 # FTLwtletters<-motif
|
|
|
357 # FTLwtmotifs[i,1]<-FTLwtletters
|
|
|
358 # FTLwtAccessionNumbers[i,1]<-FirstSubstrateSet[i,3]
|
|
|
359 # }
|
|
|
360 #
|
|
|
361 # if(YYYLettersToTheLeft>6 && YYYLettersToTheRight>6){
|
|
|
362 # motif<-YYYmotif
|
|
|
363 # #add blank spaces if the motif has less than 4 letters to the left/right
|
|
|
364 # motif<-c(leftspaces,YYYmotif,rightspaces)
|
|
|
365 # #save that motif, which is the Y and +/- 4 amino acids, including truncation
|
|
|
366 # motif<-motif[!motif %in% "x"]
|
|
|
367 # motif<-paste(motif, sep="", collapse="")
|
|
|
368 # FTLwtletters<-motif
|
|
|
369 # FTLwtmotifs[i,1]<-FTLwtletters
|
|
|
370 # FTLwtAccessionNumbers[i,1]<-FirstSubstrateSet[i,3]
|
|
|
371 #
|
|
|
372 #
|
|
|
373 # }
|
|
|
374 #
|
|
|
375 # }
|
|
|
376
|
|
|
377
|
|
|
378
|
|
|
379 # for (i in 1:nrow(SecondSubstrateSet)){
|
|
|
380 # D835letters<-SecondSubstrateSet[i,4:18]
|
|
|
381 # D835letters<-D835letters[D835letters !="XXXXX"]
|
|
|
382 # D835letters<-paste(D835letters, sep="", collapse="")
|
|
|
383 # leftspaces<-c()
|
|
|
384 # rightspaces<-c()
|
|
|
385 #
|
|
|
386 # YYYmotif <- unlist(strsplit(D835letters, split = ""))
|
|
|
387 # YYYposition <- match(x = "x", table = YYYmotif)
|
|
|
388 # #position itself tells me how much is to the left of that X by what it's number is. x at position 4 tells me that there are
|
|
|
389 # #just 3 letters to the left of x
|
|
|
390 #
|
|
|
391 # YYYLettersToTheLeft <- YYYposition - 1
|
|
|
392 # #how many letters to the right SHOULD just be length(motif)-position-1 if it's 5 long and x is at 3 then Y is at 4 and there is
|
|
|
393 # #just 1 spot to the right of Y so LettersToTheRight<-1 because 5-3-1=1
|
|
|
394 # YYYLettersToTheRight <- length(YYYmotif) - YYYposition - 1
|
|
|
395 # #then sanity check, we're currently looking only at +/-4, but this spot allows for up to +/- 7 as well, just depends on what the
|
|
|
396 # #variable the user puts in is
|
|
|
397 # if (YYYLettersToTheLeft < 7 | YYYLettersToTheRight < 7) {
|
|
|
398 # leftspaces<-rep(" ",times=(7-YYYLettersToTheLeft))
|
|
|
399 # rightspaces<-rep(" ",times=7-(YYYLettersToTheRight))
|
|
|
400 # #add blank spaces if the motif has less than 4 letters to the left/right
|
|
|
401 # motif<-c(leftspaces,YYYmotif,rightspaces)
|
|
|
402 # #save that motif, which is the Y and +/- 4 amino acids, including truncation
|
|
|
403 # motif<-motif[!motif %in% "x"]
|
|
|
404 # motif<-paste(motif, sep="", collapse="")
|
|
|
405 # D835letters<-motif
|
|
|
406 # D835Ymotifs[i,1]<-D835letters
|
|
|
407 # D835YAccessionNumbers[i,1]<-FirstSubstrateSet[i,3]
|
|
|
408 # }
|
|
|
409 #
|
|
|
410 # if(YYYLettersToTheLeft>6 && YYYLettersToTheRight>6){
|
|
|
411 # motif<-YYYmotif
|
|
|
412 # #add blank spaces if the motif has less than 4 letters to the left/right
|
|
|
413 # motif<-c(leftspaces,YYYmotif,rightspaces)
|
|
|
414 # #save that motif, which is the Y and +/- 4 amino acids, including truncation
|
|
|
415 # motif<-motif[!motif %in% "x"]
|
|
|
416 # motif<-paste(motif, sep="", collapse="")
|
|
|
417 # D835letters<-motif
|
|
|
418 # D835Ymotifs[i,1]<-D835letters
|
|
|
419 # D835YAccessionNumbers[i,1]<-FirstSubstrateSet[i,3]
|
|
|
420 # }
|
|
|
421 # }
|
|
|
422
|
|
|
423 # for (k in 1:length(FTLwtmotifsFINAL)) {
|
|
|
424 # AN<-00000
|
|
|
425 # #I don't remember why, but I felt it necessary to destroy the accession number multiple times to ensure it is
|
|
|
426 # #destroyed immediately after use
|
|
|
427 # for (m in 1:ncol(Firstsubbackfreq)) {
|
|
|
428 # AN <- as.character(Firstsubbackfreq[1, m])
|
|
|
429 # if (grepl(pattern = AN,
|
|
|
430 # x = names(FTLwtmotifsFINAL[k]),
|
|
|
431 # fixed = TRUE) == TRUE) {
|
|
|
432 # outputmatrix <- as.character(Firstsubbackfreq[, m])
|
|
|
433 # outputmatrix <- matrix(outputmatrix, nrow = 1)
|
|
|
434 # #with that accession number, find a match in the subbackfreq file and save it here
|
|
|
435 # FTLFinalMatrix<-rbind(FTLFinalMatrix,outputmatrix)
|
|
|
436 # }
|
|
|
437 # }
|
|
|
438 # }
|
|
|
439 # for (k in 1:length(D835YmotifsFINAL)) {
|
|
|
440 # #I don't remember why, but I felt it necessary to destroy the accession number multiple times to ensure it is
|
|
|
441 # #destroyed immediately after use
|
|
|
442 # for (m in 1:ncol(Secondsubbackfreq)) {
|
|
|
443 # AN <- as.character(Secondsubbackfreq[1, m])
|
|
|
444 # if (grepl(pattern = AN,
|
|
|
445 # x = names(D835YmotifsFINAL[k]),
|
|
|
446 # fixed = TRUE) == TRUE) {
|
|
|
447 # outputmatrix <- as.character(Secondsubbackfreq[, m])
|
|
|
448 # outputmatrix <- matrix(outputmatrix, nrow = 1)
|
|
|
449 # #with that accession number, find a match in the subbackfreq file and save it here
|
|
|
450 # D835YFinalMatrix<-rbind(D835YFinalMatrix,outputmatrix)
|
|
|
451 # }
|
|
|
452 # }
|
|
|
453 # } |
