Data607_Project2 - RNA_DNA

Read in data

rawDf <- read.csv("https://raw.githubusercontent.com/heathergeiger/Data607_data_to_tidy/master/RNA_vs_DNA_example_sample_sheet.csv", stringsAsFactors = FALSE, header = FALSE, colClasses = "character")

Split dataframe vertically, bind into one data set

rawDf1 <- rawDf[, 1:6]
rawDf2 <- rawDf[, 8:13]

colnames(rawDf1) <- letters[1:length(rawDf1)]
colnames(rawDf2) <- letters[1:length(rawDf2)]

fullDf <- bind_rows(rawDf1, rawDf2)

rm(rawDf, rawDf1, rawDf2)

Extract group numbers

fullDf$groupNum <- str_extract(fullDf$a, "[^0][:digit:]+") #Not sure why [^0][:digit:]+ is returning "p068", for instance
fullDf$groupNum <- str_replace(fullDf$groupNum, "p0", "")

fullDf <- fullDf %>% 
  fill(groupNum) %>%
  filter(!str_detect(a, "Group*"))

Remove unneeded rows, adjust column names

colnames(fullDf) <- fullDf[1, ]
fullDf <- fullDf %>% rename(GroupNum = `68`)

fullDf <- fullDf %>%
  filter(!(Index == "" | Index == "Index"))

Seems worth keeping group number for now, in case we wanted to compare the data sources this references in later analysis

Additional tidying to be done:

• Replace the “_" in columns “Sample” and “WGS ID” with a “-”.

fullDf <- fullDf %>% mutate_at(vars("Sample", "WGS ID"), funs(str_replace_all(., "_", "-")))

• Remove any text or underscores from WGS Quote ID to get just a five-digit number. We can count on this always being 5 digits if you want to use a regex.

fullDf$`WGS Quote ID` <- str_extract(fullDf$`WGS Quote ID`, "[:digit:]+")
stopifnot(all(str_count(fullDf$`WGS Quote ID`) == 5))

fullDf$`WGS Quote ID` <- as.numeric(fullDf$`WGS Quote ID`)

• Include an additional column that simplifies tissue type into just either “SpinalCord”,“Cerebellum”, or “Cortex”. Note that different sources may use either underscores or spaces between words in the tissue column, so best to use a regex rather than a simple match to translate these here.

fullDf$`Tissue Type` <- str_replace_all(fullDf$`Tissue Type`, "_", " ")
table(fullDf$`Tissue Type`, useNA = "ifany")

## 
##           Cerebellum       Cortex Frontal     Cortex Occipital 
##                    6                   10                    6 
## Motor Cortex Lateral  Motor Cortex Medial Spinal Cord Cervical 
##                    7                    6                    4 
##   Spinal Cord Lumbar Spinal Cord Thoracic 
##                    5                    4

• Also would be good to make sure there is only one project and gender per unique value in “WGS ID”

# Is project number "Sample" or WGS Quote Id"?  or "GroupNum"
fullDf %>% 
  group_by(`WGS ID`, Gender, Sample) %>%
  filter(n() != 1) %>% 
  nrow()

## [1] 0

fullDf %>% 
  group_by(`WGS ID`, Gender, `WGS Quote ID`) %>%
  filter(n() != 1) %>%
  nrow()

## [1] 47

Project seems to be “Sample”, which has a unique WGS ID when grouped also by Gender

• …and only unique tissues for each “WGS ID” (so no repeats of say “Spinal_Cord_Lumbar” for the same WGS ID).

fullDf %>% 
  group_by(`WGS ID`, `Tissue Type`) %>%
  filter(n() != 1) %>% 
  nrow()

## [1] 0

Tissue Type is unique by WGS ID