06-pods.Rmd

# Pods

These are all possible overlaps of pathways across all possible comparisons.

```{r}
plot = global_state$results %>%
  map( ~ pluck(., "prepared_data")) %>%
  set_names(c("GSEA", "Enrichr", "LINCS")) %>%
  bind_rows(.id = "Pod") %>%
  mutate(Group = str_replace(Group, "-C|-D", "")) %>%
  mutate(ID = paste0(Group, "_", Pod)) %>%
  select(ID, GOID) %>%
  group_by(GOID) %>%
  summarise(ID = list(ID)) %>%
  filter(!(str_detect(ID, "LINCS") & !str_detect(ID, "GSEA|Enrichr"))) %>%
  ggplot(aes(x = ID)) +
  geom_bar() +
  theme_pubr() +
  scale_x_upset() +
  labs(x = NULL, y = "Pathways")

plot

```

## 2 Pod

These are overlaps between each datasets GSEA and EnrichR results

```{r, results='asis'}
plots = global_state$results %>%
  map( ~ pluck(., "prepared_data")) %>%
  set_names(c("GSEA", "Enrichr", "LINCS")) %>%
  bind_rows(.id = "Pod") %>%
  filter(Pod != "LINCS") %>%
  group_by(Group, Pod) %>%
  summarise(GOID = list(GOID)) %>%
  group_by(Group) %>%
  nest() %>%
  transmute(Group, data = map(data, ~deframe(.))) %>%
  deframe() %>%
  map(~ ggVennDiagram(.)) %>% .[order(names(.))]

data = global_state$results %>%
  map(~ pluck(., "prepared_data")) %>%
  set_names(c("GSEA", "Enrichr", "LINCS")) %>%
  bind_rows(.id = "Pod") %>%
  filter(Pod != "LINCS") %>%
  mutate(value = ifelse(Pod == "Enrichr" | Pod == "LINCS", sign(value) * 2^(abs(value)-1), value)) %>%
  group_by(Group) %>%
  nest %>%
  deframe() %>%
  map(
    ~ list(
      `Top 10` = .x %>% 
        filter(sign(value) == 1) %>%
        group_by(GOID) %>%
        mutate(duplicated = n() > 1) %>%
        filter(duplicated) %>%
        ungroup() %>%
        select(GOID, Pod, value) %>%
        pivot_wider(names_from = "Pod", values_from = "value") %>%
        {if(ncol(.) != 1) . = arrange(., desc(GSEA), desc(Enrichr)); .} %>%
        inner_join(global_state$results$GSEA_PAVER_result$goterms_df) %>%
        select(-GOID),
      `Bottom 10` = .x %>% 
        filter(sign(value) == -1) %>%
        group_by(GOID) %>%
        mutate(duplicated = n() > 1) %>%
        filter(duplicated) %>%
        ungroup() %>%
        select(GOID, Pod, value) %>%
        pivot_wider(names_from = "Pod", values_from = "value") %>%
        {if(ncol(.) != 1) . = arrange(., GSEA, Enrichr); .} %>%
        inner_join(global_state$results$GSEA_PAVER_result$goterms_df) %>%
        select(-GOID)
    )
  ) %>% .[order(names(.))]

global_state$results <- c(global_state$results, list(`twopoddata` = data))

list(names(plots) %>% as.list(), plots, data) %>%
  pmap(~ knit_child(
    text = c(
      '',
      '### `r ..1`',
      '',
      '```{r}',
      'print(..2)',
      'make_table(..3[[1]], caption = \'Shared Upregulated Pathways\')',
      'make_table(..3[[2]], caption = \'Shared Downregulated Pathways\')',
      '```',
      ''
    ),
    envir = environment(),
    quiet = TRUE
  )) %>%
  list_c() %>%
  cat(sep = "\n")
```

## 3 Pod

These are overlaps between each datasets GSEA and EnrichR and LINCS results

```{r, results='asis'}
plots = global_state$results %>%
  map( ~ pluck(., "prepared_data")) %>%
  compact() %>%
  set_names(c("GSEA", "Enrichr", "LINCS")) %>%
  bind_rows(.id = "Pod") %>%
  mutate(Group = str_replace(Group, "-C|-D", "")) %>%
  group_by(Group, Pod) %>%
  summarise(GOID = list(GOID)) %>%
  group_by(Group) %>%
  nest() %>%
  transmute(Group, data = map(data, ~deframe(.))) %>%
  deframe() %>%
  map(~ ggVennDiagram(.)) %>% .[order(names(.))]

data = global_state$results %>%
  map(~ pluck(., "prepared_data")) %>%
  compact() %>%
  set_names(c("GSEA", "Enrichr", "LINCS")) %>%
  bind_rows(.id = "Pod") %>%
  mutate(Group = str_replace(Group, "-C|-D", "")) %>%
  mutate(value = ifelse(Pod == "Enrichr" | Pod == "LINCS", sign(value) * 2^(abs(value)-1), value)) %>%
  group_by(Group) %>%
  nest %>%
  deframe() %>%
  map(
    ~ list(
      `Top 10` = .x %>% 
        filter(sign(value) == 1) %>%
        group_by(GOID) %>%
        mutate(duplicated = n() > 2) %>%
        filter(duplicated) %>%
        ungroup() %>%
        select(GOID, Pod, value) %>%
        pivot_wider(names_from = "Pod", values_from = "value") %>%
        {if(ncol(.) != 1) . = arrange(., desc(GSEA), desc(Enrichr)); .} %>%
        inner_join(global_state$results$GSEA_PAVER_result$goterms_df) %>%
        select(-GOID),
      `Bottom 10` = .x %>% 
        filter(sign(value) == -1) %>%
        group_by(GOID) %>%
        mutate(duplicated = n() > 2) %>%
        filter(duplicated) %>%
        ungroup() %>%
        select(GOID, Pod, value) %>%
        pivot_wider(names_from = "Pod", values_from = "value") %>%
        {if(ncol(.) != 1) . = arrange(., GSEA, Enrichr); .} %>%
        inner_join(global_state$results$GSEA_PAVER_result$goterms_df) %>%
        select(-GOID)
    )
  ) %>% .[order(names(.))]

fourpoddata <- data %>% 
  map(bind_rows) %>%
  reduce(inner_join, by = "Term") %>%
  select(Term)

global_state$results <- c(global_state$results, lst(`threepoddata` = data, fourpoddata))

list(names(plots) %>% as.list(), plots, data) %>%
  pmap(~ knit_child(
    text = c(
      '',
      '### `r ..1`',
      '',
      '```{r}',
      'print(..2)',
      'make_table(..3[[1]], caption = \'Shared Upregulated Pathways\')',
      'make_table(..3[[2]], caption = \'Shared Downregulated Pathways\')',
      '```',
      ''
    ),
    envir = environment(),
    quiet = TRUE
  )) %>%
  list_c() %>%
  cat(sep = "\n")
```

<!-- ## 4th Pod -->
```{r, results='asis'}
if(nrow(global_state$results$fourpoddata) > 0) {
  global_state$results$fourpoddata %>%
    knitr::knit_child(
    text = c(
      '',
      '## 4th Pod',
      '',
      'These are the overlaps between each 3 datasets 3rd pod.',
      '',
      '```{r}',
      'make_table(.)',
      '```',
      ''
    ),
    envir = environment(),
    quiet = TRUE
  ) %>%
    cat(sep = "\n")
}
```