Efrain Gonzalez / Cleaning and Fixing Data with R

Download as
Browse Code ยป

Commit 4d40f2746596baecff36ec15a7d5b389b0aca34c

Authored by Efrain Gonzalez
1 parent 02c140cb2b
Exists in master

Update first question

Showing 1 changed file with 1 additions and 1 deletions   Show diff stats
1 #Checking for similar genes in both data sets 1 #Checking for similar genes in both data sets
2 2
3 #Required libraries 3 #Required libraries
4 library(MASS) 4 library(MASS)
5 library(readr) 5 library(readr)
6 library(dplyr) 6 library(dplyr)
7 7
8 Check2Match <- function(){ 8 Check2Match <- function(){
9 numrows <- "How many rows of clinical data are their in each data set (seperate each number by a comma no spaces)?: " %>% 9 numrows <- "How many rows of clinical data are their in each data set (separate each number by a comma no spaces)?: " %>%
10 readline(prompt = .) %>% 10 readline(prompt = .) %>%
11 strsplit(.,split = ",") %>% 11 strsplit(.,split = ",") %>%
12 .[[1]] %>% 12 .[[1]] %>%
13 as.integer(.) 13 as.integer(.)
14 i <- 1 14 i <- 1
15 ANDIS <- select.list(choices = list.files(),multiple = TRUE, title = "Choose the file/files you want to analyze:") 15 ANDIS <- select.list(choices = list.files(),multiple = TRUE, title = "Choose the file/files you want to analyze:")
16 16
17 for(i in 1:length(numrows)){ 17 for(i in 1:length(numrows)){
18 if( i == 1){ 18 if( i == 1){
19 edfile <- ANDIS[i] 19 edfile <- ANDIS[i]
20 ed <- edfile %>% 20 ed <- edfile %>%
21 read_delim(.,delim = "\t") 21 read_delim(.,delim = "\t")
22 ednocd <- ed[-(1:numrows[1]),] 22 ednocd <- ed[-(1:numrows[1]),]
23 #Second file brought in 23 #Second file brought in
24 eddfile <- ANDIS[i + 1] 24 eddfile <- ANDIS[i + 1]
25 edd <- eddfile %>% 25 edd <- eddfile %>%
26 read_delim(.,delim = "\t") 26 read_delim(.,delim = "\t")
27 eddnocd <- edd[-(1:numrows[2]),] 27 eddnocd <- edd[-(1:numrows[2]),]
28 28
29 ##Fully matched both data sets 29 ##Fully matched both data sets
30 eddy <- inner_join(ednocd,eddnocd) 30 eddy <- inner_join(ednocd,eddnocd)
31 #Matches 31 #Matches
32 meds <- eddy[,1] 32 meds <- eddy[,1]
33 33
34 } 34 }
35 if(i > 1 && i + 1 < length(numrows)){ 35 if(i > 1 && i + 1 < length(numrows)){
36 eddfile <- ANDIS[i + 1] 36 eddfile <- ANDIS[i + 1]
37 edd <- eddfile %>% 37 edd <- eddfile %>%
38 read_delim(.,delim = "\t") 38 read_delim(.,delim = "\t")
39 eddnocd <- edd[-(1:numrows[i + 1]),] 39 eddnocd <- edd[-(1:numrows[i + 1]),]
40 ##Fully matched both data sets 40 ##Fully matched both data sets
41 eddy <- inner_join(meds,eddnocd) 41 eddy <- inner_join(meds,eddnocd)
42 meds <- eddy[,1] 42 meds <- eddy[,1]
43 43
44 44
45 } 45 }
46 i <- i + 1 46 i <- i + 1
47 } 47 }
48 meds 48 meds
49 for(j in 1:length(numrows)){ 49 for(j in 1:length(numrows)){
50 edfile <- ANDIS[j] 50 edfile <- ANDIS[j]
51 ed <- edfile %>% 51 ed <- edfile %>%
52 read_delim(.,delim = "\t") 52 read_delim(.,delim = "\t")
53 ednocd <- ed[-(1:numrows[j]),] 53 ednocd <- ed[-(1:numrows[j]),]
54 #use meds to match 54 #use meds to match
55 eddy <- inner_join(meds,ednocd) 55 eddy <- inner_join(meds,ednocd)
56 Finedm <- rbind(ed[1:numrows[j],],eddy) 56 Finedm <- rbind(ed[1:numrows[j],],eddy)
57 nam_fil_ed <- strsplit(edfile,"[\\|/]") %>% 57 nam_fil_ed <- strsplit(edfile,"[\\|/]") %>%
58 .[[1]] %>% 58 .[[1]] %>%
59 .[length(.)] %>% 59 .[length(.)] %>%
60 gsub("\\D","",.) %>% 60 gsub("\\D","",.) %>%
61 c("GSE",.,"matched.txt") %>% 61 c("GSE",.,"matched.txt") %>%
62 paste(collapse = "") 62 paste(collapse = "")
63 write.table(Finedm,file = nam_fil_ed,sep = "\t",row.names = FALSE) 63 write.table(Finedm,file = nam_fil_ed,sep = "\t",row.names = FALSE)
64 j <- j + 1 64 j <- j + 1
65 } 65 }
66 meds 66 meds
67 } 67 }
68 68
69 Check2Match() 69 Check2Match()