Parkinson's curated metagenomic studies.

pMDstudies.Rmd

---
title: "pMDstudies"
author: "Levi Waldron"
date: "`r Sys.Date()`"
output: html_document
---

```{r, message=FALSE, warning=FALSE, echo=FALSE}
library(parkinsonsMetagenomicData)
library(dplyr)
library(gt)
```

```{r, echo=FALSE, warning=FALSE, message=FALSE}
pd <- select(parkinsonsMetagenomicData::sampleMetadata, c("study_name", "BioProject", "target_condition", "body_site", "host_species", "age", "age_unit", "sex"))

study_summary <- pd %>%
  group_by(study_name) %>%
  summarise(
    BioProject = paste(unique(na.omit(BioProject)), collapse = ", "),
    host_species = paste(unique(na.omit(host_species)), collapse = ", "),
    body_site = paste(unique(na.omit(body_site)), collapse = ", "),
    number_of_samples = n(),
    age_range = if(all(is.na(age))) NA_character_ else paste(min(age, na.rm = TRUE), max(age, na.rm = TRUE), sep = " - "),
    age_unit = paste(unique(na.omit(age_unit)), collapse = ", "),
    sex_distribution = paste0(
      "M: ", sum(sex == 'Male', na.rm = TRUE), " | ",
      "F: ", sum(sex == 'Female', na.rm = TRUE), " | ",
      "NA: ", sum(is.na(sex) | sex == "")
    ),
    .groups = 'drop' # Ungroup for subsequent steps
  )

gt_table <- study_summary %>%
  gt() %>%
  tab_header(
    title = md("**Study-Level Metadata Summary**"),
    subtitle = "A summary of aggregated data for each study"
  ) %>%
  cols_label(
    study_name = "Study Name",
    BioProject = "NCBI BioProject",
    host_species = "Host Species",
    body_site = "Body Site",
    number_of_samples = "Sample Count",
    age_range = "Age Range",
    age_unit = "Age Unit",
    sex_distribution = "Sex Distribution (M | F | NA)"
  ) %>%
  sub_missing(
    columns = everything(),
    missing_text = "Not Available"
  ) %>%
  tab_source_note(
    source_note = md("Data source: `sampleMetadata`")
  ) %>%
  tab_options(
    table.border.top.color = "black",
    table.border.top.width = px(3),
    column_labels.border.bottom.color = "black",
    column_labels.border.bottom.width = px(2),
    table_body.hlines.color = "#f2f2f2"
  ) %>%
  cols_align(
      align = "center",
      columns = number_of_samples
  )
gtsave(gt_table, filename = "~/Downloads/pMD_sampleMetadata.html")
```

```{r, results='asis'}
print(gt_table)
```

Output at https://rpubs.com/lwaldron/pMDstudies