Write a complete analysis to a single .nadia file

Stores the preprocessing tables, the processing results and, optionally, the Pattern Profiler output in one DuckDB database, together with the parameters and the provenance of the run. Nothing is duplicated: everything that can be derived is left out and rebuilt on reading, through SQL views stored in the file itself.

Usage

write_nadia(
  file,
  preprocessing,
  result = NULL,
  pattern_profiler = NULL,
  overwrite = FALSE,
  verbose = TRUE
)

Arguments

file: Path to write to. The .nadia extension is a convention; the file is a DuckDB database whatever it is called.
preprocessing: A proteomics_data object, from any of the preprocess_*() functions. Required: it is the base of the file.
result: Object from process_proteomics(), or NULL to store only the preprocessing.
pattern_profiler: Object from pattern_profiler_analysis(), or NULL. Can also be added later with nadia_add_pattern_profiler().
overwrite: Replace file if it already exists.
verbose: Print progress.

Value

The path, invisibly.

Details

Use read_nadia() to read it back, and nadia_result() to get an object the rest of the package can plot directly.

What goes in

Twelve tables. samples and proteins are the dimensions; protein_sample_metrics, intensities, imputed_values and de_results the facts; pp_membership and pp_profile the optional clustering; and nadia_meta, nadia_packages, nadia_source_files, nadia_calls, parameters and nadia_schema the provenance.

Three things are deliberately not stored, because they are exactly derivable: the log2 assay (log2 of the raw one), the PCA_Input table (the imputed assay joined to the DE results) and the observed cells of the imputed assay (identical to the normalized ones). The last of these is checked rather than assumed, since a model-based method such as limpa can revise observed values; when it does, the full assay is stored instead.

Where not to put it

Not inside a synchronised folder such as iCloud Drive, Dropbox or OneDrive while you are working. DuckDB keeps the file open with locks and partial writes, and a synchronising client can corrupt it. Write it locally and copy it afterwards.

Examples

if (requireNamespace("duckdb", quietly = TRUE)) {
    data(nadia_dia)
    res <- process_proteomics(nadia_dia, verbose = FALSE)

    f <- file.path(tempdir(), "example.nadia")
    write_nadia(f, nadia_dia, res, verbose = FALSE)

    db <- read_nadia(f)
    db

    unlink(f)
}

Write a complete analysis to a single `.nadia` file