create_meta.Rmd

---
title: "Creating the metadata"
author: "Mark Dunning"
date: "26/07/2018"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

## Required libraries

```{r}
library(GEOquery)
library(dplyr)
library(SRAdb)
```

## Getting meta-data from GEO

```{r}
tmp <- getGEO("GSE60450")
gseInf <- pData(tmp[[1]])
gseInf
```

## Getting meta-data from SRA

```{r}
sqlfile <-'SRAmetadb.sqlite'
if(!file.exists('SRAmetadb.sqlite')) sqlfile <<- getSRAdbFile()
sra_con <- dbConnect(SQLite(),sqlfile)

sraInf <- getSRAinfo("SRP045534",sra_con, sraType="sra")
sraInf
```


## Joining the meta-data

```{r}
gseInf <- mutate(gseInf, experiment = substr(as.character(relation.1),44,52))
gseInf
dir.create("meta_data",showWarnings = FALSE)

combinedInf <- left_join(gseInf, sraInf, by="experiment")

combinedInf <- rename(combinedInf,CellType=`immunophenotype:ch1`,Status=`developmental stage:ch1`) %>% 
  mutate(CellType = ifelse(grepl("luminal",CellType),"luminal","basal")) %>% 
  mutate(Status = gsub("18.5 day ", "", Status)) %>% 
  mutate(Status = gsub("2 day ","",Status)) %>% 
  mutate(Name = basename(as.character(supplementary_file_1))) %>% 
  mutate(Name = substr(Name, 12,18)) %>% 
  select(run, Name,CellType,Status)


write.table(combinedInf, file="meta_data/sampleInfo_Corrected.txt",sep="\t",row.names=FALSE)

## Create a sample swap

combinedInf$CellType[combinedInf$Name == "MCL1-DH"] <- "luminal"
combinedInf$CellType[combinedInf$Name == "MCL1-LA"] <- "basal"

write.table(combinedInf, file="meta_data/sampleInfo.txt",sep="\t",row.names=FALSE)


combinedInf$CellType[c(3,4)] <- stringr:::str_to_title(as.character(combinedInf$CellType[c(3,4)]))
combinedInf$Status[c(11,12)] <- paste0(" ", as.character(combinedInf$Status[c(11,12)]))
```