3a_ImmunoSEQ_setup_metadata_and_loadData.Rmd

---
title: "Metadata creation"
author: "Daniel Shu"
date: "`r Sys.Date()`"
output: html_document
---

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

# I. Load libraries

```{r}
library(tidyverse)
library(readxl)
library(kableExtra)

data_dir = "data/adaptive/"
output.path = "output/adaptive/"
#creates output.path directory if not already present
ifelse(!dir.exists(output.path), dir.create(output.path), paste0(output.path, " ", "directory already exists"))
```

# II. Create common metadata object for all samples 

```{r}
#create metadata dataframe using adaptive filenames
df = data.frame(Sample = list.files(path=data_dir, pattern=".tsv") %>% str_replace(".tsv", "")) 
df$Patient.ID = df$Sample %>% str_replace("-.*|_PBMC.*|_Biopsy.*", "")
df$time_point = if_else(grepl(("Pre|Biopsy"), df$Sample), "Pre", "Post") 
df$repertoire = ifelse(endsWith(df$Sample, "TCRB"), "TCRB", "IGH")
df$type = if_else(grepl("Biopsy", df$Sample), "Biopsy", 
                  if_else(grepl("PBMC", df$Sample), "PBMC",
                          if_else(grepl("TDLN", df$Sample), "TDLN", "TLS")))
df$label = df$Sample %>% gsub("-T_TCRB$|_TCRB$|-B$", "", .) 
```

### b. Add histological metadata based on CD20 and CD21 staining

```{r}
meta_hist = read_delim("data/metadata_histology_pub.txt", show_col_types = FALSE)
meta_hist=rename(meta_hist, label = TLS) #rename TLS column to 'label', which will be used for joining, 

df_hist = df[df$type=="TLS",] %>% select(.,label) %>% unique()
df_hist= left_join(df_hist, meta_hist, by='label')

#add df_hist
df = left_join(df,df_hist, by='label')
```

### c. Add clinical metadata from Bulk RNAseq analysis

```{r}
metadata_clinical <- readRDS("output/sampInfo.rds") 
#subset for neoadjuvant treated
metadata_clinical = metadata_clinical[metadata_clinical$group=="Neoadjuvant",]

 #subset out metadata that I don't need to include in analysis
desired <- c('Publication.ID', 'Age', 'Sex', 'Treatment', 'Response', 'Response.2', 'Relapse', 'Death', 'Etiology', 'Prior.HBV', 'Prior.HCV', 'Histologic.grade')
metadata_clinical <- metadata_clinical %>% select(all_of(desired))
metadata_clinical
colnames(metadata_clinical) == desired #check that we got columns we wanted 

#change metadata_clinical to remove 0s and 1s
metadata_clinical$Sex <- 
  factor(metadata_clinical$Sex, levels=c(1,0),
         labels=c("Male", 
                  "Female"))
metadata_clinical$Relapse <- 
  factor(metadata_clinical$Relapse, levels=c("Yes","No"),
         labels=c("Relapse", 
                  "No_relapse"))
metadata_clinical$Death <- 
  factor(metadata_clinical$Death, levels=c(1,0),
         labels=c("Dead", 
                  "Alive"))
metadata_clinical$Prior.HBV <- 
  factor(metadata_clinical$Prior.HBV, levels=c(1,0),
         labels=c("Prior_HBV", 
                  "No_Prior_HBV"))
metadata_clinical$Prior.HCV <- 
  factor(metadata_clinical$Prior.HCV, levels=c(1,0),
         labels=c("Prior_HCV", 
                  "No_Prior_HCV"))

#join to df
df = left_join(df, metadata_clinical, by = c("Patient.ID"= "Publication.ID"))
```

### d. create myLabels object and add columns with publication labels
```{r}
#add consecutive group number, code from https://statisticsglobe.com/assign-unique-id-for-each-group-in-r
df$Location = factor(df$Location, levels=c("Intratumoral", "Peritumoral", "Adjacent Normal"))
df$TLS_type = factor(df$TLS_type, levels=c("Mature", "Involuted"))


myLabels = select(df, "label", "Patient.ID", "TLS_type", "Location") %>% distinct()
myLabels = myLabels %>% arrange(.,Patient.ID, TLS_type, Location, label) %>%
  group_by(Patient.ID) %>%

# add Order column for use with immunarch
  mutate(Order=row_number())
 
# #reverse order column for pre and post PBMC (due to alphabetical ordering, the pre-treatment is currently n+1 compared to posttreatment, I want the pre-treatment to have n-1 posttreatment)
myLabels[grep("PBMC_Pre", myLabels$label),]$Order =  myLabels[grep("PBMC_Pre", myLabels$label),]$Order-1
myLabels[grep("PBMC_Post", myLabels$label),]$Order =  myLabels[grep("PBMC_Post", myLabels$label),]$Order+1
# 
myLabels = arrange(myLabels, Patient.ID, Order)

df = merge(df,myLabels[,c("label", "Order")], by="label")

df = arrange(df, Patient.ID, Order)
```

### e. export final metadata file to data folder

```{r}
write.table(df, file = paste0(data_dir,"metadata.txt"), quote=F, sep = "\t",
            row.names = F, col.names = T)
```

# III. SessionInfo

```{r}
sessionInfo()
writeLines(capture.output(sessionInfo()), "sessionInfo_ImmunoSEQ_setup_metadata_and_loadData.txt")
```