Mercurial > repos > jfb > difference_finder
comparison all stuff/Difference finderMADE 7 TO 7 1-15-2019.R @ 0:23eea82f5192 draft
Uploaded
author | jfb |
---|---|
date | Wed, 16 Jan 2019 13:55:22 -0500 |
parents | |
children | 110a74dff8e1 |
comparison
equal
deleted
inserted
replaced
-1:000000000000 | 0:23eea82f5192 |
---|---|
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 | |
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("S1.csv", stringsAsFactors=FALSE) | |
16 Firstsubbackfreq<- read.csv("SBF1.csv", header=FALSE, stringsAsFactors=FALSE) | |
17 | |
18 SecondSubstrateSet<- read.csv("S2.csv", stringsAsFactors=FALSE) | |
19 Secondsubbackfreq<- read.csv("SBF2.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<-"1R1 substrates.csv" | |
32 First_unshared_subbackfreq<-"1R1 SBF.csv" | |
33 | |
34 Second_unshared_motifs_table<-"1R2 subs.csv" | |
35 Second_unshared_subbackfreq<-"1R2 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(FirstCentralLetters, grepl, pattern="S", ignore.case=TRUE) | |
58 SecondTees<-sapply(FirstCentralLetters, grepl, pattern="T", ignore.case=TRUE) | |
59 SecondWys<-sapply(FirstCentralLetters, 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,FirstEsses,"xS") | |
66 SecondCentralLetters<-replace(SecondCentralLetters,FirstTees,"xT") | |
67 SecondCentralLetters<-replace(SecondCentralLetters,FirstWys,"xY") | |
68 | |
69 #################################################################################################################################### | |
70 #################################################################################################################################### | |
71 # better version of this code written in C: what happens when two kinases share a motif, but they found that motif in two | |
72 # separate proteins thus two separate accession numbers? | |
73 # It should actually output the shared motif and BOTH accession numbers. Right now it does not, it only maps out the second | |
74 # accession number. So that needs to be fixed BUT you need to keep the commonality between a motif and its accession number | |
75 #################################################################################################################################### | |
76 #################################################################################################################################### | |
77 #################################################################################################################################### | |
78 #################################################################################################################################### | |
79 | |
80 #Create the motif sets, deciding wether or not you're looking for truncated or full here | |
81 #full only | |
82 | |
83 | |
84 if (Are_You_Looking_For_Commonality=="NO"){ | |
85 | |
86 ############################################### | |
87 #ALL motifs, full and truncated | |
88 | |
89 if (FullMotifsOnly_questionmark!="YES"&&TruncatedMotifsOnly_questionmark!="YES"){ | |
90 FTLwtmotifs=matrix(,nrow = nrow(FirstSubstrateSet),ncol=1) | |
91 FTLwtAccessionNumbers=matrix(,nrow = nrow(FirstSubstrateSet),ncol=1) | |
92 | |
93 for (i in 1:nrow(FirstSubstrateSet)){ | |
94 FTLwtletters<-FirstSubstrateSet[i,4:18] | |
95 FTLwtletters<-FTLwtletters[FTLwtletters !="XXXXX"] | |
96 FTLwtletters<-paste(FTLwtletters, sep="", collapse="") | |
97 leftspaces<-c() | |
98 rightspaces<-c() | |
99 | |
100 YYYmotif <- unlist(strsplit(FTLwtletters, split = "")) | |
101 YYYposition <- match(x = "x", table = YYYmotif) | |
102 #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 | |
103 #just 3 letters to the left of x | |
104 | |
105 YYYLettersToTheLeft <- YYYposition - 1 | |
106 #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 | |
107 #just 1 spot to the right of Y so LettersToTheRight<-1 because 5-3-1=1 | |
108 YYYLettersToTheRight <- length(YYYmotif) - YYYposition - 1 | |
109 #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 | |
110 #variable the user puts in is | |
111 | |
112 | |
113 if (YYYLettersToTheLeft < 7 | YYYLettersToTheRight < 7) { | |
114 leftspaces<-rep(" ",times=(7-YYYLettersToTheLeft)) | |
115 rightspaces<-rep(" ",times=7-(YYYLettersToTheRight)) | |
116 #add blank spaces if the motif has less than 4 letters to the left/right | |
117 motif<-c(leftspaces,YYYmotif,rightspaces) | |
118 #save that motif, which is the Y and +/- 4 amino acids, including truncation | |
119 motif<-motif[!motif %in% "x"] | |
120 motif<-paste(motif, sep="", collapse="") | |
121 FTLwtletters<-motif | |
122 FTLwtmotifs[i,1]<-FTLwtletters | |
123 FTLwtAccessionNumbers[i,1]<-FirstSubstrateSet[i,3] | |
124 } | |
125 | |
126 if(YYYLettersToTheLeft>6 && YYYLettersToTheRight>6){ | |
127 motif<-YYYmotif | |
128 #add blank spaces if the motif has less than 4 letters to the left/right | |
129 motif<-c(leftspaces,YYYmotif,rightspaces) | |
130 #save that motif, which is the Y and +/- 4 amino acids, including truncation | |
131 motif<-motif[!motif %in% "x"] | |
132 motif<-paste(motif, sep="", collapse="") | |
133 FTLwtletters<-motif | |
134 FTLwtmotifs[i,1]<-FTLwtletters | |
135 FTLwtAccessionNumbers[i,1]<-FirstSubstrateSet[i,3] | |
136 | |
137 | |
138 } | |
139 | |
140 } | |
141 | |
142 D835Ymotifs=matrix(,nrow = nrow(SecondSubstrateSet),ncol=1) | |
143 D835YAccessionNumbers<-matrix(,nrow = nrow(SecondSubstrateSet),ncol = 1) | |
144 | |
145 for (i in 1:nrow(SecondSubstrateSet)){ | |
146 D835letters<-SecondSubstrateSet[i,4:18] | |
147 D835letters<-D835letters[D835letters !="XXXXX"] | |
148 D835letters<-paste(D835letters, sep="", collapse="") | |
149 leftspaces<-c() | |
150 rightspaces<-c() | |
151 | |
152 YYYmotif <- unlist(strsplit(D835letters, split = "")) | |
153 YYYposition <- match(x = "x", table = YYYmotif) | |
154 #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 | |
155 #just 3 letters to the left of x | |
156 | |
157 YYYLettersToTheLeft <- YYYposition - 1 | |
158 #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 | |
159 #just 1 spot to the right of Y so LettersToTheRight<-1 because 5-3-1=1 | |
160 YYYLettersToTheRight <- length(YYYmotif) - YYYposition - 1 | |
161 #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 | |
162 #variable the user puts in is | |
163 if (YYYLettersToTheLeft < 7 | YYYLettersToTheRight < 7) { | |
164 leftspaces<-rep(" ",times=(7-YYYLettersToTheLeft)) | |
165 rightspaces<-rep(" ",times=7-(YYYLettersToTheRight)) | |
166 #add blank spaces if the motif has less than 4 letters to the left/right | |
167 motif<-c(leftspaces,YYYmotif,rightspaces) | |
168 #save that motif, which is the Y and +/- 4 amino acids, including truncation | |
169 motif<-motif[!motif %in% "x"] | |
170 motif<-paste(motif, sep="", collapse="") | |
171 D835letters<-motif | |
172 D835Ymotifs[i,1]<-D835letters | |
173 D835YAccessionNumbers[i,1]<-FirstSubstrateSet[i,3] | |
174 } | |
175 | |
176 if(YYYLettersToTheLeft>6 && YYYLettersToTheRight>6){ | |
177 motif<-YYYmotif | |
178 #add blank spaces if the motif has less than 4 letters to the left/right | |
179 motif<-c(leftspaces,YYYmotif,rightspaces) | |
180 #save that motif, which is the Y and +/- 4 amino acids, including truncation | |
181 motif<-motif[!motif %in% "x"] | |
182 motif<-paste(motif, sep="", collapse="") | |
183 D835letters<-motif | |
184 D835Ymotifs[i,1]<-D835letters | |
185 D835YAccessionNumbers[i,1]<-FirstSubstrateSet[i,3] | |
186 } | |
187 } | |
188 | |
189 names(FTLwtmotifs)<-FTLwtAccessionNumbers | |
190 names(D835Ymotifs)<-D835YAccessionNumbers | |
191 | |
192 } | |
193 | |
194 | |
195 FTLwtmotifsFINAL<-FTLwtmotifs[!FTLwtmotifs %in% D835Ymotifs] | |
196 FTLwtmotifsFINAL<-FTLwtmotifsFINAL[!duplicated(FTLwtmotifsFINAL)] | |
197 | |
198 D835YmotifsFINAL<-D835Ymotifs[!D835Ymotifs %in% FTLwtmotifs] | |
199 D835YmotifsFINAL<-D835YmotifsFINAL[!duplicated(D835YmotifsFINAL)] | |
200 | |
201 | |
202 columnalheader<-c(rep(NA,36)) | |
203 FTLFinalMatrix<-matrix(data =columnalheader,nrow = 1) | |
204 | |
205 for (k in 1:length(FTLwtmotifsFINAL)) { | |
206 AN<-00000 | |
207 #I don't remember why, but I felt it necessary to destroy the accession number multiple times to ensure it is | |
208 #destroyed immediately after use | |
209 for (m in 1:ncol(Firstsubbackfreq)) { | |
210 AN <- as.character(Firstsubbackfreq[1, m]) | |
211 if (grepl(pattern = AN, | |
212 x = names(FTLwtmotifsFINAL[k]), | |
213 fixed = TRUE) == TRUE) { | |
214 outputmatrix <- as.character(Firstsubbackfreq[, m]) | |
215 outputmatrix <- matrix(outputmatrix, nrow = 1) | |
216 #with that accession number, find a match in the subbackfreq file and save it here | |
217 FTLFinalMatrix<-rbind(FTLFinalMatrix,outputmatrix) | |
218 } | |
219 } | |
220 } | |
221 FTLFinalMatrix<-FTLFinalMatrix[!duplicated(FTLFinalMatrix),] | |
222 FTLFinalMatrix<-FTLFinalMatrix[2:nrow(FTLFinalMatrix),] | |
223 | |
224 columnalheader<-c(rep(NA,36)) | |
225 D835YFinalMatrix<-matrix(data =columnalheader,nrow = 1) | |
226 | |
227 for (k in 1:length(D835YmotifsFINAL)) { | |
228 #I don't remember why, but I felt it necessary to destroy the accession number multiple times to ensure it is | |
229 #destroyed immediately after use | |
230 for (m in 1:ncol(Secondsubbackfreq)) { | |
231 AN <- as.character(Secondsubbackfreq[1, m]) | |
232 if (grepl(pattern = AN, | |
233 x = names(D835YmotifsFINAL[k]), | |
234 fixed = TRUE) == TRUE) { | |
235 outputmatrix <- as.character(Secondsubbackfreq[, m]) | |
236 outputmatrix <- matrix(outputmatrix, nrow = 1) | |
237 #with that accession number, find a match in the subbackfreq file and save it here | |
238 D835YFinalMatrix<-rbind(D835YFinalMatrix,outputmatrix) | |
239 } | |
240 } | |
241 } | |
242 D835YFinalMatrix<-D835YFinalMatrix[!duplicated(D835YFinalMatrix),] | |
243 D835YFinalMatrix<-D835YFinalMatrix[2:nrow(D835YFinalMatrix),] | |
244 | |
245 FTLoutputmatrix<-matrix(data=c(FTLwtmotifsFINAL,names(FTLwtmotifsFINAL)),ncol = 2) | |
246 | |
247 | |
248 FLTheader<-c("Substrate","Species","Reference","-7","-6","-5","-4","-3","-2","-1","0","1","2","3","4","5","6","7","Phosphite") | |
249 # FLTheader<-unlist(FLTheader) | |
250 lefthandFLT<-matrix(data = rep(NA,times=2*nrow(FTLoutputmatrix)),nrow=nrow(FTLoutputmatrix)) | |
251 righthandFLT<-matrix(data = rep(NA,times=1*nrow(FTLoutputmatrix)),nrow=nrow(FTLoutputmatrix)) | |
252 FLTaccessionset<-FTLoutputmatrix[,2] | |
253 FTLmeat<-sapply(FTLoutputmatrix[,1], strsplit, "") | |
254 FTLmeat<-sapply(FTLmeat, unlist) | |
255 colnames(FTLmeat)<-NULL | |
256 FTLmeat<-t(FTLmeat) | |
257 | |
258 FTLoutputmatrix2<-cbind(lefthandFLT,FLTaccessionset,FTLmeat,righthandFLT) | |
259 colnames(FTLoutputmatrix2)<-NULL | |
260 rownames(FTLoutputmatrix2)<-NULL | |
261 colnames(FLTheader)<-NULL | |
262 rownames(FLTheader)<-NULL | |
263 | |
264 | |
265 FirstCentralLettersAGAIN<-FTLoutputmatrix2[,11] | |
266 | |
267 FirstEsses<-sapply(FirstCentralLettersAGAIN, grepl, pattern="S", ignore.case=TRUE) | |
268 FirstTees<-sapply(FirstCentralLettersAGAIN, grepl, pattern="T", ignore.case=TRUE) | |
269 FirstWys<-sapply(FirstCentralLettersAGAIN, grepl, pattern="Y", ignore.case=TRUE) | |
270 | |
271 FirstCentralLettersAGAIN<-replace(FirstCentralLettersAGAIN,FirstEsses,"xS") | |
272 FirstCentralLettersAGAIN<-replace(FirstCentralLettersAGAIN,FirstTees,"xT") | |
273 FirstCentralLettersAGAIN<-replace(FirstCentralLettersAGAIN,FirstWys,"xY") | |
274 | |
275 FirstCentralLettersAGAIN->FTLoutputmatrix2[,11] | |
276 | |
277 FTLoutputmatrix2<-rbind(FLTheader,FTLoutputmatrix2) | |
278 | |
279 write.table(x=FTLoutputmatrix2, | |
280 file=First_unshared_motifs_table, | |
281 quote=FALSE, sep=",", | |
282 row.names=FALSE,col.names = FALSE, na="", append=TRUE) | |
283 | |
284 columnalheader<-c("Accession Numbers",as.character(Firstsubbackfreq[1:35,1])) | |
285 columnalheader<-matrix(columnalheader,nrow = 1) | |
286 write.table(x=columnalheader, | |
287 file=First_unshared_subbackfreq, | |
288 quote=FALSE, sep=",", | |
289 row.names=FALSE,col.names = FALSE, na="", append=TRUE) | |
290 | |
291 write.table(x=FTLFinalMatrix, | |
292 file=First_unshared_subbackfreq, | |
293 quote=FALSE, sep=",", | |
294 row.names=FALSE,col.names = FALSE, na="", append=TRUE) | |
295 | |
296 ############################################################################################################ | |
297 | |
298 D835Youtputmatrix<-matrix(data=c(D835YmotifsFINAL,names(D835YmotifsFINAL)),ncol = 2) | |
299 | |
300 D835Yheader<-c("Substrate","Species","Reference","-7","-6","-5","-4","-3","-2","-1","0","1","2","3","4","5","6","7","Phosphite") | |
301 # D835Yheader<-unlist(D835Yheader) | |
302 lefthandD835<-matrix(data = rep(NA,times=2*nrow(D835Youtputmatrix)),nrow=nrow(D835Youtputmatrix)) | |
303 righthandD835<-matrix(data = rep(NA,times=1*nrow(D835Youtputmatrix)),nrow=nrow(D835Youtputmatrix)) | |
304 D835Yaset<-D835Youtputmatrix[,2] | |
305 D835meat<-sapply(D835Youtputmatrix[,1], strsplit, "") | |
306 D835meat<-sapply(D835meat, unlist) | |
307 colnames(D835meat)<-NULL | |
308 D835meat<-t(D835meat) | |
309 | |
310 D835Youtputmatrix2<-cbind(lefthandD835,D835Yaset,D835meat,righthandD835) | |
311 colnames(D835Youtputmatrix2)<-NULL | |
312 rownames(D835Youtputmatrix2)<-NULL | |
313 colnames(D835Yheader)<-NULL | |
314 rownames(D835Yheader)<-NULL | |
315 | |
316 | |
317 SecondCentralLettersAGAIN<-D835Youtputmatrix2[,11] | |
318 | |
319 SecondEsses<-sapply(SecondCentralLettersAGAIN, grepl, pattern="S", ignore.case=TRUE) | |
320 SecondTees<-sapply(SecondCentralLettersAGAIN, grepl, pattern="T", ignore.case=TRUE) | |
321 SecondWys<-sapply(SecondCentralLettersAGAIN, grepl, pattern="Y", ignore.case=TRUE) | |
322 | |
323 SecondCentralLettersAGAIN<-replace(SecondCentralLettersAGAIN,FirstEsses,"xS") | |
324 SecondCentralLettersAGAIN<-replace(SecondCentralLettersAGAIN,FirstTees,"xT") | |
325 SecondCentralLettersAGAIN<-replace(SecondCentralLettersAGAIN,FirstWys,"xY") | |
326 | |
327 SecondCentralLettersAGAIN->D835Youtputmatrix2[,11] | |
328 | |
329 D835Youtputmatrix2<-rbind(D835Yheader,D835Youtputmatrix2) | |
330 | |
331 write.table(x=D835Youtputmatrix2, | |
332 file=Second_unshared_motifs_table, | |
333 quote=FALSE, sep=",", | |
334 row.names=FALSE,col.names = FALSE, na="", append=TRUE) | |
335 | |
336 columnalheader<-c("Accession Numbers",as.character(Firstsubbackfreq[1:35,1])) | |
337 columnalheader<-matrix(columnalheader,nrow = 1) | |
338 write.table(x=columnalheader, | |
339 file=Second_unshared_subbackfreq, | |
340 quote=FALSE, sep=",", | |
341 row.names=FALSE,col.names = FALSE, na="", append=TRUE) | |
342 | |
343 write.table(x=D835YFinalMatrix, | |
344 file=Second_unshared_subbackfreq, | |
345 quote=FALSE, sep=",", | |
346 row.names=FALSE,col.names = FALSE, na="", append=TRUE) | |
347 } | |
348 |