Generate targets pipeline code for GWAS meta-analysis

Generates a complete targets / tarchetypes pipeline as a character string, ready to be pasted into a _targets.R file. The generated code covers per-cohort preparation, LDSC heritability estimation, per-ancestry and all-populations IVW meta-analysis, loci extraction, Manhattan plots (with PDF and high-DPI PNG export targets for fast report rendering), and MR-MEGA MDS. Validates trait_type, n_col, and manifest_df before generating any code. The manifest is serialized as a tibble::tribble() call in the generated output so the pipeline is self-contained.

Usage

generate_gwas_meta_pipeline(
  trait,
  trait_type,
  n_col,
  manifest_df,
  hm3_path,
  dbsnp_path,
  crew_controller = NULL,
  output_base_dir = "Data",
  manhattan_width = 16,
  manhattan_height = 6,
  manhattan_dpi = 300,
  output_file = NULL
)

Arguments

trait

Trait name used in target names (e.g. "CAD").

trait_type

"binary" or "quantitative" (required, no default).

n_col

Sample size column: "EffectiveN" for binary traits, "N" for quantitative traits (required, no default).

manifest_df

A data frame describing the cohort files to include. Required columns: path (full path to raw summary statistics file), file (basename of the file), cohort (cohort label), ancestry (ancestry label, e.g. "EUR"), study (unique human-readable identifier; must contain no spaces or special characters; does not need to include the ancestry). A tar_name column ({study}_{ancestry}) is automatically added and used as the tar_map target name so that ancestry-based target selectors work reliably. Additional columns are preserved in the serialized output.

Optionally, the manifest may include col_* columns to map non-standard column names in each cohort's raw file to harmonized names. These are passed to build_column_map() during per-cohort preparation. Supported columns: col_chr, col_pos, col_rsid, col_effect_allele, col_other_allele, col_beta, col_se, col_p, col_eaf, col_n, col_n_cases, col_n_controls. See build_column_map() for details.

hm3_path

Path to the LDSC HapMap3 SNP list file (e.g. w_hm3.snplist). Required.

dbsnp_path

Path to the dbSNP155 reference directory passed to prep_gwas(). Required.

crew_controller

Name of the crew controller to use for resource-intensive targets (meta_ALL and meta_common_variants_ALL). If NULL (default), no resources block is added to those targets.

output_base_dir

Base output directory for intermediate files. Default "Data".

manhattan_width

Width in inches for Manhattan plot export. Default 16.

manhattan_height

Height in inches for Manhattan plot export. Default 6.

manhattan_dpi

DPI for high-resolution Manhattan plot PNG export. Default 300.

output_file

Optional file path to write the generated code to. When provided, parent directories are created automatically, the code is wrapped in a ```{targets <trait>-pipeline} chunk, and the raw pipeline code string is returned invisibly. When NULL (default), the string is returned visibly.

Value

A length-1 character string containing valid R code defining a targets pipeline.

Details

Generate a targets GWAS meta-analysis pipeline

Examples

if (FALSE) { # \dontrun{
manifest <- data.frame(
  path     = c("/data/UKBB_EUR.txt.gz", "/data/MVP_AFR.txt.gz"),
  file     = c("UKBB_EUR.txt.gz",       "MVP_AFR.txt.gz"),
  cohort   = c("UKBB",                  "MVP"),
  ancestry = c("EUR",                   "AFR"),
  study    = c("UKBB_EUR",              "MVP_AFR"),
  # Optional: map non-standard column names per cohort
  col_eaf  = c("MY_FREQ",              NA),
  col_beta = c("BETA_VAL",             NA),
  stringsAsFactors = FALSE
)
cat(generate_gwas_meta_pipeline("CAD", trait_type = "binary",
                                n_col = "EffectiveN", manifest_df = manifest,
                                hm3_path   = "/path/to/w_hm3.snplist",
                                dbsnp_path = "/path/to/dbSNP155"))
} # }