Skip to contents

Validate a Single Study's Data Files Against the Project Schema

validate_study.Rd

Reads the metadata.yml and data.csv files for a specific study ID within a metawoRld project and checks their structure and content against the rules defined in the project's _metawoRld.yml schema. It also checks internal consistency (e.g., links between data and metadata).

This is useful for checking manually edited files (e.g., those created using add_study_template).

Usage

validate_study(study_id, path = ".", check_linkages = TRUE)

Arguments

study_id

Character string. The unique identifier of the study to validate.

path

Character string. The path to the root directory of the metawoRld project. Defaults to the current working directory (.).

check_linkages

Logical. Should the consistency checks between data.csv (method_ref_id, group_label) and metadata.yml (measurement_methods, outcome_groups) be performed? Defaults to TRUE.

Value

Returns TRUE (invisibly) if all validation checks pass. If any check fails, the function stops with an informative error message using rlang::abort().

Examples

if (FALSE) { # \dontrun{
# --- Setup: Create a project and add a template ---
proj_path <- file.path(tempdir(), "validate_study_test")
create_metawoRld(
  path = proj_path,
  project_name = "Validate Study Test",
  project_description = "Testing validate_study()"
)
add_study_template(proj_path, "ManualStudy01")

# --- Scenario 1: Validate the raw template (might fail required field checks) ---
# This will likely fail because placeholders like "REQUIRED: ..." are still present.
tryCatch(
  validate_study("ManualStudy01", path = proj_path),
  error = function(e) print(paste("Validation Failed (as expected for raw template):", e$message))
)

# --- Scenario 2: Manually Edit Files (Simulated) ---
# Imagine the user fills the files. Let's simulate correct filling:
meta_file <- file.path(proj_path, "data", "ManualStudy01", "metadata.yml")
data_file <- file.path(proj_path, "data", "ManualStudy01", "data.csv")

# Create valid-looking content programmatically for the example
valid_meta <- list(
  study_id = "ManualStudy01", title = "Manually Entered Study",
  authors = list("User U"), year = 2024, journal = "Data Entry Journal",
  study_design = "Cross-sectional", country = "Local", sample_type = "Serum",
  outcome_groups = list(
    g1 = list(name = "Group A", definition = "Criteria A"),
    g2 = list(name = "Group B", definition = "Criteria B")
  ),
  measurement_methods = list(
    m_elisa = list(analysis_type = "ELISA", target_cytokine = "CYTOK", unit = "pg/mL")
  )
  # Optional fields omitted for brevity
)
yaml::write_yaml(valid_meta, meta_file)

valid_data <- data.frame(
  measurement_id = "meas1", method_ref_id = "m_elisa", cytokine_name = "CYTOK",
  group_label = "g1", gestational_age_timing = "Any", n = 50,
  statistic_type = "mean_sd", value1 = 10.0, value2 = 2.5
)
readr::write_csv(valid_data, data_file)

# --- Scenario 3: Validate the correctly filled study ---
validate_study("ManualStudy01", path = proj_path) # Should now pass and print success

# --- Scenario 4: Introduce an error (e.g., bad group_label in data.csv) ---
invalid_data <- valid_data
invalid_data$group_label[1] <- "g_BAD" # This label doesn't exist in metadata
readr::write_csv(invalid_data, data_file)

tryCatch(
  validate_study("ManualStudy01", path = proj_path),
  error = function(e) print(paste("Validation Failed (invalid group link):", e$message))
)

# --- Scenario 5: Validate without checking linkages ---
# This should pass even with the bad group_label, as linkage check is skipped
validate_study("ManualStudy01", path = proj_path, check_linkages = FALSE)

# --- Clean up ---
unlink(proj_path, recursive = TRUE)
} # }