Fixed issue with similar variable names

Efrain Gonzalez
1 parent 8f1c6201bb
Showing 1 changed file with 36 additions and 17 deletions Show diff stats
RMarkovBlanket.r
@@ -76,7 +76,7 @@ for(i in 1:sizeDotP1){
 			NewDotP2_2[chngreq,1] <- gsub(paste0("^",NewDotP1[i,2],"$"),NewDotP1[i,1],NewDotP2_2[chngreq,1])
 		}
 	}
-	i <- i + 1
+	#i <- i + 1
 }
 NewDotP2_2
 for(j in 1:sizeDotP1){
@@ -87,11 +87,11 @@ for(j in 1:sizeDotP1){
 			NewDotP2_2[chngreq,2] <- gsub(paste0("^",NewDotP1[j,2],"$"),NewDotP1[j,1],NewDotP2_2[chngreq,2])
 		}
 	}
-	j <- j + 1
+	#j <- j + 1
 }
  
 lrgMarkov <- dim(NewDotP2_2)[1]
-Blanky <- function(MarkovDegree = lrgMarkov, VariableStartName = "Alzheimer", VariableEndName = "Age"){
+Blanky <- function(MarkovDegree = 20, VariableStartName = "Alzheimer", VariableEndName = "GRIN2A"){
 	#Finding the Parents and Children
 	d <- 1
 	AllNamList1 <- vector("list",length = 3)
@@ -100,20 +100,25 @@ Blanky &lt;- function(MarkovDegree = lrgMarkov, VariableStartName = &quot;Alzheimer&quot;, Va
 	##A list of lists
 	##outer set by the degree of the Markov blanket
 	AllVarList <- vector("list",length = 3)
+	varfound <- 0
 	for(d in 1:MarkovDegree){
 		colnames(NewDotP2_2) <- NULL
 		##Which variable are you looking for?
 		##This is the VariableEndName
 		if(d == 1){
 			##Finding the Parents for the variable
-			LocPofVar <- grep(VariableStartName,NewDotP2_2[,2])
+			LocPofVar <- grep(paste0("^",VariableStartName,"$"),NewDotP2_2[,2])
 			PofVar <- NewDotP2_2[LocPofVar,1]
 			AllNamList1[[1]] <- PofVar
+			AllNamList1[[1]] <- AllNamList1[[1]][!duplicated(AllNamList1[[1]])]
+			varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllNamList1[[1]]))
  
 			##Finding the Children for the variable
-			LocCofVar <- grep(VariableStartName,NewDotP2_2[,1])
+			LocCofVar <- grep(paste0("^",VariableStartName,"$"),NewDotP2_2[,1])
 			CofVar <- NewDotP2_2[LocCofVar,2]
 			AllNamList1[[2]] <- CofVar
+			AllNamList1[[2]] <- AllNamList1[[2]][!duplicated(AllNamList1[[2]])]
+			varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllNamList1[[2]]))
  
 			##Finding the Co-Parents of the Children for the variable
 			NumofChild <- length(CofVar)
@@ -123,25 +128,29 @@ Blanky &lt;- function(MarkovDegree = lrgMarkov, VariableStartName = &quot;Alzheimer&quot;, Va
 				COPlist <- vector("character", length = 0)
 				nc <- 1
 				for(nc in 1:NumofChild){
-					LocCOPofVar <- grep(CofVar[nc],NewDotP2_2[,2])
+					LocCOPofVar <- grep(paste0("^",CofVar[nc],"$"),NewDotP2_2[,2])
 					COPofVar <- NewDotP2_2[LocCOPofVar,1]	
 					if(sum(grepl(VariableStartName,COPofVar)) >= 1){
 						#positions of variable start name within the vector of co parents
-						posoforig <- grep(VariableStartName,COPofVar)
+						posoforig <- grep(paste0("^",VariableStartName,"$"),COPofVar)
 						COPofVar <- COPofVar[-posoforig]
 						COPlist <- append(COPlist,COPofVar)
 					} else{
 						#COPlist[[nc]] <- COPofVar[COPofVar!=VariableStartName]
 						COPlist <- append(COPlist,COPofVar)
 					}
-					nc <- nc + 1
+					#nc <- nc + 1
 				} 
 			} else {
 				##Making COPlist empty
 				COPlist <- vector("character",length = 0)
 			}
 			AllNamList1[[3]] <- COPlist
+			AllNamList1[[3]] <- AllNamList1[[3]][!duplicated(AllNamList1[[3]])]
+			varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllNamList1[[3]]))
+			
 			AllVarList[[1]] <- AllNamList1
+
 		} else if(d > 1){
 			##inner set by the length of the previous AllVarlist we are working on
 			lPreVList <- length(AllVarList[[d-1]])
@@ -162,16 +171,19 @@ Blanky &lt;- function(MarkovDegree = lrgMarkov, VariableStartName = &quot;Alzheimer&quot;, Va
 					PofVlist <- vector("character", length = 0)
 					np <- 1
 					for(np in 1:NumofVars){
-						LocPofVar <- grep(AllVarList[[d-1]][[ef]][np],NewDotP2_2[,2])
+						LocPofVar <- grep(paste0("^",AllVarList[[d-1]][[ef]][np],"$"),NewDotP2_2[,2])
 						PofVar <- NewDotP2_2[LocPofVar,1]
 						PofVlist <- append(PofVlist,PofVar)
-						np <- np + 1
+						#np <- np + 1
 					}
 				} else {
 					##Making COPlist empty
 					PofVlist <- vector("character",length = 0)
 				}
 				AllVarList[[d]][[PCCP]] <- PofVlist 
+				AllVarList[[d]][[PCCP]] <- AllVarList[[d]][[PCCP]][!duplicated(AllVarList[[d]][[PCCP]])]
+				##Have you found the VariableEndName?
+				varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllVarList[[d]][[PCCP]]))
 				PCCP <- PCCP + 1
  
 				##Finding the Children
@@ -181,7 +193,7 @@ Blanky &lt;- function(MarkovDegree = lrgMarkov, VariableStartName = &quot;Alzheimer&quot;, Va
 					CofVlist <- vector("character", length = 0)
 					np <- 1
 					for(np in 1:NumofVars){
-						LocCofVar <- grep(AllVarList[[d-1]][[ef]][np],NewDotP2_2[,1])
+						LocCofVar <- grep(paste0("^",AllVarList[[d-1]][[ef]][np],"$"),NewDotP2_2[,1])
 						CofVar <- NewDotP2_2[LocCofVar,2]
 						#if(sum(grepl(VariableStartName,CofVar)) >= 1){
 						#	#positions of variable start name within the vector of co parents
@@ -191,13 +203,16 @@ Blanky &lt;- function(MarkovDegree = lrgMarkov, VariableStartName = &quot;Alzheimer&quot;, Va
 						#} else{
 							CofVlist <- append(CofVlist,CofVar)
 						#}
-						np <- np + 1
+						#np <- np + 1
 					}
 				} else {
 					##Making CofPlist empty
 					CofVlist <- vector("character",length = 0)
 				}
 				AllVarList[[d]][[PCCP]] <- CofVlist
+				AllVarList[[d]][[PCCP]] <- AllVarList[[d]][[PCCP]][!duplicated(AllVarList[[d]][[PCCP]])]
+				##Have you found the VariableEndName yet?
+				varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllVarList[[d]][[PCCP]]))
 				PCCP <- PCCP + 1
  
 				##Finding the Co-Parents
@@ -206,7 +221,7 @@ Blanky &lt;- function(MarkovDegree = lrgMarkov, VariableStartName = &quot;Alzheimer&quot;, Va
 					ncp <- 1
 					CPofClist <- vector("character",length = 0)
 					for(ncp in 1:NumofCVars){
-						LocCPofCVar <- grep(CofVlist[ncp],NewDotP2_2[,2])
+						LocCPofCVar <- grep(paste0("^",CofVlist[ncp],"$"),NewDotP2_2[,2])
 						CPofCVar <- NewDotP2_2[LocCPofCVar,1]
 						#if(sum(grepl(,CPofCVar)) >= 1){
 						#	#positions of variable start name within the vector of co parents
@@ -216,7 +231,7 @@ Blanky &lt;- function(MarkovDegree = lrgMarkov, VariableStartName = &quot;Alzheimer&quot;, Va
 						#} else{
 							CPofClist <- append(CPofClist,CPofCVar)
 						#}
-						ncp <- ncp + 1
+						#ncp <- ncp + 1
 					}
  
 				} else {
@@ -224,15 +239,19 @@ Blanky &lt;- function(MarkovDegree = lrgMarkov, VariableStartName = &quot;Alzheimer&quot;, Va
 					CPofClist <- vector("character",length = 0)
 				}
 				AllVarList[[d]][[PCCP]] <- CPofClist
+				AllVarList[[d]][[PCCP]] <- AllVarList[[d]][[PCCP]][!duplicated(AllVarList[[d]][[PCCP]])]
+				##Have you found VariableEndName now?
+				varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllVarList[[d]][[PCCP]]))
 				PCCP <- PCCP + 1		
-				ef <- ef + 1
+				#ef <- ef + 1
 			}
 		}
+
 		##Stop if you have found the VariableEndName value
-		if(sum(grepl(VariableEndName,AllVarList)) > 0){
+		if(varfound > 0){
 			break
 		}
-		d <- d + 1
+		#d <- d + 1
 	}
 	##The Markov Degree is that found below
 	d
...	...	@@ -76,7 +76,7 @@ for(i in 1:sizeDotP1){
76	76	NewDotP2_2[chngreq,1] <- gsub(paste0("^",NewDotP1[i,2],"$"),NewDotP1[i,1],NewDotP2_2[chngreq,1])
77	77	}
78	78	}
79		- i <- i + 1
	79	+ #i <- i + 1
80	80	}
81	81	NewDotP2_2
82	82	for(j in 1:sizeDotP1){
...	...	@@ -87,11 +87,11 @@ for(j in 1:sizeDotP1){
87	87	NewDotP2_2[chngreq,2] <- gsub(paste0("^",NewDotP1[j,2],"$"),NewDotP1[j,1],NewDotP2_2[chngreq,2])
88	88	}
89	89	}
90		- j <- j + 1
	90	+ #j <- j + 1
91	91	}
92	92
93	93	lrgMarkov <- dim(NewDotP2_2)[1]
94		-Blanky <- function(MarkovDegree = lrgMarkov, VariableStartName = "Alzheimer", VariableEndName = "Age"){
	94	+Blanky <- function(MarkovDegree = 20, VariableStartName = "Alzheimer", VariableEndName = "GRIN2A"){
95	95	#Finding the Parents and Children
96	96	d <- 1
97	97	AllNamList1 <- vector("list",length = 3)
...	...	@@ -100,20 +100,25 @@ Blanky <- function(MarkovDegree = lrgMarkov, VariableStartName = "Alzheimer", Va
100	100	##A list of lists
101	101	##outer set by the degree of the Markov blanket
102	102	AllVarList <- vector("list",length = 3)
	103	+ varfound <- 0
103	104	for(d in 1:MarkovDegree){
104	105	colnames(NewDotP2_2) <- NULL
105	106	##Which variable are you looking for?
106	107	##This is the VariableEndName
107	108	if(d == 1){
108	109	##Finding the Parents for the variable
109		- LocPofVar <- grep(VariableStartName,NewDotP2_2[,2])
	110	+ LocPofVar <- grep(paste0("^",VariableStartName,"$"),NewDotP2_2[,2])
110	111	PofVar <- NewDotP2_2[LocPofVar,1]
111	112	AllNamList1[[1]] <- PofVar
	113	+ AllNamList1[[1]] <- AllNamList1[[1]][!duplicated(AllNamList1[[1]])]
	114	+ varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllNamList1[[1]]))
112	115
113	116	##Finding the Children for the variable
114		- LocCofVar <- grep(VariableStartName,NewDotP2_2[,1])
	117	+ LocCofVar <- grep(paste0("^",VariableStartName,"$"),NewDotP2_2[,1])
115	118	CofVar <- NewDotP2_2[LocCofVar,2]
116	119	AllNamList1[[2]] <- CofVar
	120	+ AllNamList1[[2]] <- AllNamList1[[2]][!duplicated(AllNamList1[[2]])]
	121	+ varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllNamList1[[2]]))
117	122
118	123	##Finding the Co-Parents of the Children for the variable
119	124	NumofChild <- length(CofVar)
...	...	@@ -123,25 +128,29 @@ Blanky <- function(MarkovDegree = lrgMarkov, VariableStartName = "Alzheimer", Va
123	128	COPlist <- vector("character", length = 0)
124	129	nc <- 1
125	130	for(nc in 1:NumofChild){
126		- LocCOPofVar <- grep(CofVar[nc],NewDotP2_2[,2])
	131	+ LocCOPofVar <- grep(paste0("^",CofVar[nc],"$"),NewDotP2_2[,2])
127	132	COPofVar <- NewDotP2_2[LocCOPofVar,1]
128	133	if(sum(grepl(VariableStartName,COPofVar)) >= 1){
129	134	#positions of variable start name within the vector of co parents
130		- posoforig <- grep(VariableStartName,COPofVar)
	135	+ posoforig <- grep(paste0("^",VariableStartName,"$"),COPofVar)
131	136	COPofVar <- COPofVar[-posoforig]
132	137	COPlist <- append(COPlist,COPofVar)
133	138	} else{
134	139	#COPlist[[nc]] <- COPofVar[COPofVar!=VariableStartName]
135	140	COPlist <- append(COPlist,COPofVar)
136	141	}
137		- nc <- nc + 1
	142	+ #nc <- nc + 1
138	143	}
139	144	} else {
140	145	##Making COPlist empty
141	146	COPlist <- vector("character",length = 0)
142	147	}
143	148	AllNamList1[[3]] <- COPlist
	149	+ AllNamList1[[3]] <- AllNamList1[[3]][!duplicated(AllNamList1[[3]])]
	150	+ varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllNamList1[[3]]))
	151	+
144	152	AllVarList[[1]] <- AllNamList1
	153	+
145	154	} else if(d > 1){
146	155	##inner set by the length of the previous AllVarlist we are working on
147	156	lPreVList <- length(AllVarList[[d-1]])
...	...	@@ -162,16 +171,19 @@ Blanky <- function(MarkovDegree = lrgMarkov, VariableStartName = "Alzheimer", Va
162	171	PofVlist <- vector("character", length = 0)
163	172	np <- 1
164	173	for(np in 1:NumofVars){
165		- LocPofVar <- grep(AllVarList[[d-1]][[ef]][np],NewDotP2_2[,2])
	174	+ LocPofVar <- grep(paste0("^",AllVarList[[d-1]][[ef]][np],"$"),NewDotP2_2[,2])
166	175	PofVar <- NewDotP2_2[LocPofVar,1]
167	176	PofVlist <- append(PofVlist,PofVar)
168		- np <- np + 1
	177	+ #np <- np + 1
169	178	}
170	179	} else {
171	180	##Making COPlist empty
172	181	PofVlist <- vector("character",length = 0)
173	182	}
174	183	AllVarList[[d]][[PCCP]] <- PofVlist
	184	+ AllVarList[[d]][[PCCP]] <- AllVarList[[d]][[PCCP]][!duplicated(AllVarList[[d]][[PCCP]])]
	185	+ ##Have you found the VariableEndName?
	186	+ varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllVarList[[d]][[PCCP]]))
175	187	PCCP <- PCCP + 1
176	188
177	189	##Finding the Children
...	...	@@ -181,7 +193,7 @@ Blanky <- function(MarkovDegree = lrgMarkov, VariableStartName = "Alzheimer", Va
181	193	CofVlist <- vector("character", length = 0)
182	194	np <- 1
183	195	for(np in 1:NumofVars){
184		- LocCofVar <- grep(AllVarList[[d-1]][[ef]][np],NewDotP2_2[,1])
	196	+ LocCofVar <- grep(paste0("^",AllVarList[[d-1]][[ef]][np],"$"),NewDotP2_2[,1])
185	197	CofVar <- NewDotP2_2[LocCofVar,2]
186	198	#if(sum(grepl(VariableStartName,CofVar)) >= 1){
187	199	# #positions of variable start name within the vector of co parents
...	...	@@ -191,13 +203,16 @@ Blanky <- function(MarkovDegree = lrgMarkov, VariableStartName = "Alzheimer", Va
191	203	#} else{
192	204	CofVlist <- append(CofVlist,CofVar)
193	205	#}
194		- np <- np + 1
	206	+ #np <- np + 1
195	207	}
196	208	} else {
197	209	##Making CofPlist empty
198	210	CofVlist <- vector("character",length = 0)
199	211	}
200	212	AllVarList[[d]][[PCCP]] <- CofVlist
	213	+ AllVarList[[d]][[PCCP]] <- AllVarList[[d]][[PCCP]][!duplicated(AllVarList[[d]][[PCCP]])]
	214	+ ##Have you found the VariableEndName yet?
	215	+ varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllVarList[[d]][[PCCP]]))
201	216	PCCP <- PCCP + 1
202	217
203	218	##Finding the Co-Parents
...	...	@@ -206,7 +221,7 @@ Blanky <- function(MarkovDegree = lrgMarkov, VariableStartName = "Alzheimer", Va
206	221	ncp <- 1
207	222	CPofClist <- vector("character",length = 0)
208	223	for(ncp in 1:NumofCVars){
209		- LocCPofCVar <- grep(CofVlist[ncp],NewDotP2_2[,2])
	224	+ LocCPofCVar <- grep(paste0("^",CofVlist[ncp],"$"),NewDotP2_2[,2])
210	225	CPofCVar <- NewDotP2_2[LocCPofCVar,1]
211	226	#if(sum(grepl(,CPofCVar)) >= 1){
212	227	# #positions of variable start name within the vector of co parents
...	...	@@ -216,7 +231,7 @@ Blanky <- function(MarkovDegree = lrgMarkov, VariableStartName = "Alzheimer", Va
216	231	#} else{
217	232	CPofClist <- append(CPofClist,CPofCVar)
218	233	#}
219		- ncp <- ncp + 1
	234	+ #ncp <- ncp + 1
220	235	}
221	236
222	237	} else {
...	...	@@ -224,15 +239,19 @@ Blanky <- function(MarkovDegree = lrgMarkov, VariableStartName = "Alzheimer", Va
224	239	CPofClist <- vector("character",length = 0)
225	240	}
226	241	AllVarList[[d]][[PCCP]] <- CPofClist
	242	+ AllVarList[[d]][[PCCP]] <- AllVarList[[d]][[PCCP]][!duplicated(AllVarList[[d]][[PCCP]])]
	243	+ ##Have you found VariableEndName now?
	244	+ varfound <- varfound + sum(grepl(paste0("^",VariableEndName,"$"),AllVarList[[d]][[PCCP]]))
227	245	PCCP <- PCCP + 1
228		- ef <- ef + 1
	246	+ #ef <- ef + 1
229	247	}
230	248	}
	249	+
231	250	##Stop if you have found the VariableEndName value
232		- if(sum(grepl(VariableEndName,AllVarList)) > 0){
	251	+ if(varfound > 0){
233	252	break
234	253	}
235		- d <- d + 1
	254	+ #d <- d + 1
236	255	}
237	256	##The Markov Degree is that found below
238	257	d