02-geo-upload.Rmd

---
title: "Upload data to GEO"
author: "Tu Hu"
date: "06/07/2022"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(dplyr)
library(tidyr)
library(purrr)
library(stringr)
library(SummarizedExperiment)
```

## Upload data to GEO

### md5 check sum
```{r checksum files, eval=FALSE}
md5sum <- 
  readr::read_delim("data/geo/geo_md5sum.txt", delim = "  ", 
                  col_names = c("md5sum", "file_name")) %>% 
  mutate(file_name=file_name %>% str_remove("read/"),
         lib_id = file_name %>% str_extract("lib\\d{1,}"))
```

### sample metadata
```{r metadata for GEO, eval=FALSE}
se <- readr::read_rds("data/se.rds")
geo_metadata <- 
  colData(se) %>% as_tibble() %>% 
  mutate(title = paste(visit, subject, skin_type, replicate_ID,
                       sep = "_"),
         `source name` = "skin biopsy",
         organism = "Homo sapians") %>% 
  select(BAM_ID,
         title, 
         `source name`,
         organism,
         subject,
         skin_type,
         gender,
         scorad,
         easi_total_score,
         visit, date_visit,
         biopsy_area,
         library_ID)

geo_metadata_n <-
  geo_metadata %>% 
  # select(title, library_ID) %>%
  left_join(md5sum, by = c("library_ID" = "lib_id")) %>% 
  group_by(title) %>% nest() %>% 
  mutate(raw_file = data %>% map("file_name"),
         md5sum = data %>% map("md5sum"),
         data = data %>% map(~ select(.x, -md5sum, -file_name))) %>% 
  unnest(cols = data) %>% 
  distinct()
```

```{r eval=FALSE}
geo_metadata_n %>% 
  openxlsx::write.xlsx("data/geo/geo_meta.xlsx")
```

### count matrix
```{r} 
geo_md5sum <- 
  geo_metadata_n %>% 
  ungroup %>% 
  unnest(cols = c(raw_file, md5sum)) %>% 
  select(raw_file, md5sum)
openxlsx::write.xlsx(geo_md5sum, "data/geo/geo_md5sum.xlsx")

count_matrix <-
  assay(se, 1) %>% as_tibble(rownames = "gene_name")
colnames(count_matrix)[-1] <- geo_metadata$title
readr::write_csv(count_matrix, "data/geo/count_matrix.csv")

count_matrix_normalized <-
  assay(se, 2) %>% as_tibble(rownames = "gene_name")
colnames(count_matrix_normalized)[-1] <- geo_metadata$title
readr::write_csv(count_matrix_normalized, "data/geo/count_matrix_normalized.csv")
```

### paired-end data specification
```{r paired, eval=FALSE}
geo_md5sum %>% 
  select(raw_file) %>% 
  mutate(raw_file = raw_file %>% str_remove("_(1|2).fastq.gz")) %>% 
  distinct() %>% 
  mutate(pair_1 = paste0(raw_file, "_1.fastq.gz"),
         pair_2 = paste0(raw_file, "_2.fastq.gz")) %>% 
  select(contains("pair")) %>% 
  openxlsx::write.xlsx("data/geo/geo_paired.xlsx")
```