00-BIOMAP.qmd

---
title: "Biomap federated analysis"
author: "Tu Hu"
date: "06/07/2022"
output: html_document
---

```{r}
library(dplyr)
library(gprofiler2)
```

```{r}
count_matrix <- 
  readr::read_csv("data/geo/count_matrix.csv")

gene_id <- 
  count_matrix$gene_name %>% 
  gprofiler2::gconvert() %>% 
  as_tibble()

gene_id_filt <- 
  gene_id %>% 
  group_by(input) %>% 
  tidyr::nest() %>% 
  mutate(n = purrr::map_int(data, nrow)) %>% 
  filter(n == 1) %>% 
  tidyr::unnest() %>% 
  dplyr::select(input, target)

count_matrix_filt <- 
  count_matrix %>% 
  filter(gene_name %in% gene_id_filt$input) %>% 
  left_join(gene_id_filt, by = c("gene_name" = "input")) %>% 
  dplyr::select(gene_id = target, contains("_")) %>% 
  dplyr::select(-gene_name) %>% 
  dplyr::select(gene_id, contains(c("LS"))) %>% 
  # dplyr::select(gene_id, starts_with("01")) %>% 
  distinct(gene_id, .keep_all = T)

# readr::write_csv(count_matrix_filt_bl, "data/biomap_count_matrix_filt_all_visit.csv")
```

```{r}
phenotype <- 
  openxlsx::read.xlsx("data/geo/geo_meta.xlsx") %>% 
  as_tibble()

meta_data_manuscript <- 
  openxlsx::read.xlsx("data/supplementary/table_s1.xlsx") %>% 
  as_tibble()

phenotype <-
  tibble(
    title = colnames(count_matrix_filt_bl)[-1]
  ) %>% 
  left_join(phenotype) %>% 
  select(id = title, 
         baseline_characteristics_sex = gender, 
         subject, skin_type,
         scorad, easi_total_score, visit, date_visit, biopsy_area) %>% 
  left_join(meta_data_manuscript) %>% 
  rename(
    # baseline_characteristics_age = age,
         baseline_disease_activity_SCORAD = scorad,
         baseline_disease_activity_EASI = easi_total_score) %>% 
  mutate(baseline_characteristics_sex= factor(baseline_characteristics_sex),
         baseline_characteristics_drug_washout = factor("no"))


age_age_onset <- 
  phenotype %>% filter(visit == "01") %>% select(subject, age, age_onset) %>% distinct() %>% arrange(subject) %>% 
  mutate(baseline_characteristics_onset_age = case_when(
    age_onset == "<1" ~ "0",
    TRUE ~ age_onset
  ),
  baseline_characteristics_age = age,
  baseline_characteristics_onset_age = as.numeric(baseline_characteristics_onset_age)) %>% 
  select(-age_onset)

phenotype_age_onset <- 
  phenotype %>% left_join(age_age_onset, by = "subject") %>% 
  mutate(baseline_characteristics_onset_early_ad = ifelse(baseline_characteristics_onset_age <=2, "yes", "no"),
        baseline_characteristics_onset_late_ad = ifelse(baseline_characteristics_onset_age > 18, "yes", "no")) %>% 
  select(id, visit, contains("baseline"))
```

```{r}
# file.copy("data/biomap_phenotype_bl.csv", "../adpso_endotypes_federated/data/biomap_phenotype_bl.csv", overwrite = T)
# file.copy("data/biomap_count_matrix_filt_bl.csv", "../adpso_endotypes_federated/data/biomap_count_matrix_filt_bl.csv", overwrite = T)


# v1
count_v1 <- count_matrix_filt %>% select(gene_id, starts_with("01_"))
meta_v1 <- phenotype_age_onset %>% filter(visit == "01") %>% select(-visit)

readr::write_csv(count_v1, "data/biomap_count_v1.csv")
readr::write_csv(meta_v1, "data/biomap_meta_v1.csv")

# V2
count_v2 <- count_matrix_filt %>% select(gene_id, starts_with("02_"))
meta_v2 <- phenotype_age_onset %>% filter(visit == "02") %>% select(-visit)

readr::write_csv(count_v2, "data/biomap_count_v2.csv")
readr::write_csv(meta_v2, "data/biomap_meta_v2.csv")

# V3
count_v3 <- count_matrix_filt %>% select(gene_id, starts_with("03_"))
meta_v3 <- phenotype_age_onset %>% filter(visit == "03")%>% select(-visit)

readr::write_csv(count_v3, "data/biomap_count_v3.csv")
readr::write_csv(meta_v3, "data/biomap_meta_v3.csv")

# V4
count_v4 <- count_matrix_filt %>% select(gene_id, starts_with("04_"))
meta_v4 <- phenotype_age_onset %>% filter(visit == "04")%>% select(-visit)

readr::write_csv(count_v4, "data/biomap_count_v4.csv")
readr::write_csv(meta_v4, "data/biomap_meta_v4.csv")

# V5
count_v5 <- count_matrix_filt %>% select(gene_id, starts_with("05_"))
meta_v5 <- phenotype_age_onset %>% filter(visit == "05")%>% select(-visit)

readr::write_csv(count_v5, "data/biomap_count_v5.csv")
readr::write_csv(meta_v5, "data/biomap_meta_v5.csv")

# all visit
count_all_visit <- count_matrix_filt
meta_all_visit <- phenotype_age_onset%>% select(-visit)
readr::write_csv(count_all_visit, "data/biomap_count_all_visit.csv")
readr::write_csv(meta_all_visit, "data/biomap_meta_all_visit.csv")
```

```{bash}
cp data/biomap* ../adpso_endotypes_federated/input
```