Skip to contents

Get the provenance of fits stored in a SummarizedExperiment

Source: R/fit_provenance.R
fit_provenance.Rd

apply_fit and apply_fits record a recipe for every fit in metadata(se)$provenance. These functions read it back.

Usage

get_SE_provenance(se)

get_fit_provenance(se, assay, fit_source, data_source = NULL)

summarize_fit_provenance(se)

Arguments

se

SummarizedExperiment

assay

string; name of the assay holding the fit results.

fit_source

string; value of the fit_source column.

data_source

string; for an SE built by merge_SE, the name of the merged object the rows come from (the data_source column). NULL reads the fits done on se itself.

Value

get_SE_provenance(): the whole provenance entry, or NULL for objects without one. get_fit_provenance(): one recipe. summarize_fit_provenance(): a data.table with one row per recorded fit.

Details

A recipe contains:

fn, summary_fn

record of the function: name, package, version, git_sha and hash for a function from an installed package; source, captured (values the function uses from its environment), helpers (source of helper functions it calls) and dependencies (versions of the packages it calls) for any other function; info set with set_fit_fn_info; env for declared runtime dependencies.

fit_args

arguments passed to the fit function.

called_by, data_type, input_assay, output_assay, summary_assay, slicing_cols, slicing_values, merge

settings of the call.

run

R and gDRutils versions, hash of the input assay and the number of rows written.

Names listed under not_stored or args_not_stored were too large or not plain data, so the fit cannot be fully reproduced from the object alone.

For a summary assay, get_fit_provenance() returns the recipe of the fit that wrote it. For an SE built by merge_SE, the recipes of the merged objects are kept per data_source.

Examples

mae <- get_synthetic_data("finalMAE_small.qs2")
se <- mae[["single-agent"]]
se <- apply_fit(
  se, fit_drug_response_metrics, "single-agent",
  output_assay = "custom_hill", fit_source = "hill_loose",
  fit_args = list(pcutoff = 0.1)
)
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
#> Warning: NaNs produced
summarize_fit_provenance(se)
#>          assay fit_source                                            fn
#>         <char>     <char>                                        <char>
#> 1: custom_hill hill_loose gDRutils::fit_drug_response_metrics (1.11.18)
#>         fit_args
#>           <char>
#> 1: pcutoff = 0.1
#>                                                                model
#>                                                               <char>
#> 1: 3-parameter log-logistic (drc::LL.3u), upper asymptote fixed at 1
get_fit_provenance(se, "custom_hill", "hill_loose")$fit_args
#> $pcutoff
#> [1] 0.1
#>