For matching genes between two consolidated data files (UNTESTED)

Efrain Gonzalez
1 parent 2bf073a859
Showing 1 changed file with 62 additions and 0 deletions Show diff stats
RMatchGenes.R
@@ -0,0 +1,62 @@
+#Checking for similar genes in both data sets
+
+#Required libraries
+library(MASS)
+library(readr)
+library(dplyr)
+
+Check2Match <- function(){
+#Bring in the two files
+##Number of rows with clinical data for first file
+numrow1 <- "How many rows of clinical data are their in the first data set?: " %>%
+	readline(prompt = .) %>%
+	as.integer(.)
+
+##Number of rows with clinical data for second file
+numrow2 <- "How many rows of clinical data are their in the second data set?: " %>%
+	readline(prompt = .) %>%
+	as.integer(.)
+
+edfile <- file.choose()
+ed <- edfile %>%
+	read_delim(.,delim = "\t")
+
+ednocd <- ed[-(1:numrow1),]
+eddfile <- file.choose()
+edd <- eddfile %>%
+	read_delim(.,delim = "\t")
+	
+eddnocd <- edd[-(1:numrow2),]
+
+##Number of columns that belong to the first data file
+numbcol1 <- dim(ednocd)[2]
+##Number of columns that belong to the second data file
+numbcol2 <- dim(eddnocd)[2]
+
+##Fully matched both data sets
+eddy <- inner_join(ednocd,eddnocd)
+#Matched ed
+eddy[,1:numbcol1]
+Finedm <- rbind(ed[1:numrow1,],eddy[,1:numbcol1])
+nam_fil_ed <- strsplit(edfile,"[\\|/]") %>%
+			.[[1]] %>%
+			.[length(.)] %>%
+			gsub("\\D","",.) %>%
+			c("GSE",.,"matched.txt") %>%
+			paste(collapse = "")
+write.table(Finedm,file = nam_fil_ed,sep = "\t",row.names = FALSE)
+#Matched edd
+eddy[,(numbcol1 + 1):dim(eddy)[2]]
+fineddm <- cbind(eddy[,1],eddy[,(numbcol1 + 1):dim(eddy)[2]])
+Fineddm <- rbind(edd[1:numrow2,],fineddm)
+nam_fil_edd <- strsplit(eddfile,"[\\|/]") %>%
+			.[[1]] %>%
+			.[length(.)] %>%
+			gsub("\\D","",.) %>%
+			c("GSE",.,"matched.txt") %>%
+			paste(collapse = "")
+write.table(Fineddm,file = nam_fil_edd,sep = "\t",row.names = FALSE)
+
+}
+
+Check2Match()
...	...	@@ -0,0 +1,62 @@
	1	+#Checking for similar genes in both data sets
	2	+
	3	+#Required libraries
	4	+library(MASS)
	5	+library(readr)
	6	+library(dplyr)
	7	+
	8	+Check2Match <- function(){
	9	+#Bring in the two files
	10	+##Number of rows with clinical data for first file
	11	+numrow1 <- "How many rows of clinical data are their in the first data set?: " %>%
	12	+ readline(prompt = .) %>%
	13	+ as.integer(.)
	14	+
	15	+##Number of rows with clinical data for second file
	16	+numrow2 <- "How many rows of clinical data are their in the second data set?: " %>%
	17	+ readline(prompt = .) %>%
	18	+ as.integer(.)
	19	+
	20	+edfile <- file.choose()
	21	+ed <- edfile %>%
	22	+ read_delim(.,delim = "\t")
	23	+
	24	+ednocd <- ed[-(1:numrow1),]
	25	+eddfile <- file.choose()
	26	+edd <- eddfile %>%
	27	+ read_delim(.,delim = "\t")
	28	+
	29	+eddnocd <- edd[-(1:numrow2),]
	30	+
	31	+##Number of columns that belong to the first data file
	32	+numbcol1 <- dim(ednocd)[2]
	33	+##Number of columns that belong to the second data file
	34	+numbcol2 <- dim(eddnocd)[2]
	35	+
	36	+##Fully matched both data sets
	37	+eddy <- inner_join(ednocd,eddnocd)
	38	+#Matched ed
	39	+eddy[,1:numbcol1]
	40	+Finedm <- rbind(ed[1:numrow1,],eddy[,1:numbcol1])
	41	+nam_fil_ed <- strsplit(edfile,"[\\\|/]") %>%
	42	+ .[[1]] %>%
	43	+ .[length(.)] %>%
	44	+ gsub("\\D","",.) %>%
	45	+ c("GSE",.,"matched.txt") %>%
	46	+ paste(collapse = "")
	47	+write.table(Finedm,file = nam_fil_ed,sep = "\t",row.names = FALSE)
	48	+#Matched edd
	49	+eddy[,(numbcol1 + 1):dim(eddy)[2]]
	50	+fineddm <- cbind(eddy[,1],eddy[,(numbcol1 + 1):dim(eddy)[2]])
	51	+Fineddm <- rbind(edd[1:numrow2,],fineddm)
	52	+nam_fil_edd <- strsplit(eddfile,"[\\\|/]") %>%
	53	+ .[[1]] %>%
	54	+ .[length(.)] %>%
	55	+ gsub("\\D","",.) %>%
	56	+ c("GSE",.,"matched.txt") %>%
	57	+ paste(collapse = "")
	58	+write.table(Fineddm,file = nam_fil_edd,sep = "\t",row.names = FALSE)
	59	+
	60	+}
	61	+
	62	+Check2Match()