Skip to contents

Contrast Configuration

Source: R/ContrastConfiguration.R
ContrastConfiguration.Rd

Holds the column-role mapping and behaviour flags that describe how a specific contrast backend (lm, limma, SAINT, ...) lays out its result table. Mirrors AnalysisConfiguration for the modelling side: it carries names, not data.

Consumers that need to operate on a contrast table generically (filter, build ORA inputs, build rank lists, summarise per protein) should read column names through the accessors on this object rather than hard-coding them. This lets backends with different column conventions (e.g. SAINT's BFDR/log2_EFCs) plug into the same downstream code path as LM-style backends without renaming columns.

Defaults match the standard prolfqua contrast schema produced by Contrasts, ContrastsLimma, ContrastsModerated etc.

Value

An R6 class generator.

See also

Other configuration: AnalysisConfiguration, INTERNAL_FUNCTIONS_BY_FAMILY, R6_extract_values(), complete_cases(), make_interaction_column(), make_reduced_hierarchy_config(), sample_subset(), separate_hierarchy(), setup_analysis(), table_factors(), table_factors_size()

Public fields

subject_id

columns identifying the contrast subject (e.g. protein id)

model_name_col

column carrying the model name label

contrast_col

column carrying the contrast label

effect_col

column with the (signed) effect size; for LM this is diff, for SAINT log2_EFCs

score_col

column with the per-contrast test statistic

pvalue_col

column with the raw p-value, or NA_character_ for backends that do not produce one (e.g. SAINTexpress)

fdr_col

column with the FDR / BFDR / adjusted p-value

avg_abundance_col

column with the per-feature mean abundance

supports_dea_qc

whether the backend supports the legacy differential-expression QC HTML report

needs_saint_annotation

whether the backend needs annotation reading in SAINT mode (bait / control columns)

significance_directional

if TRUE, the backend only considers positive-effect features biologically meaningful (e.g. SAINTexpress, where negative log2_EFCs are not interpreted as "down" hits). Affects filter_significant(): when TRUE the filter is one-sided on the effect column; when FALSE it is symmetric (|effect| > threshold).

Methods

Public methods

Method new()

Construct a ContrastConfiguration. All arguments map directly to fields of the same name.

Usage

ContrastConfiguration$new(
  subject_id = character(),
  model_name_col = "modelName",
  contrast_col = "contrast",
  effect_col = "diff",
  score_col = "statistic",
  pvalue_col = "p.value",
  fdr_col = "FDR",
  avg_abundance_col = "avgAbd",
  supports_dea_qc = TRUE,
  needs_saint_annotation = FALSE,
  significance_directional = FALSE
)

Arguments

subject_id

columns identifying the contrast subject

model_name_col

model-name column

contrast_col

contrast-label column

effect_col

signed effect column

score_col

test-statistic column

pvalue_col

raw p-value column, or NA_character_

fdr_col

FDR / adjusted-p column

avg_abundance_col

mean-abundance column

supports_dea_qc

supports DEA QC HTML

needs_saint_annotation

needs SAINT-mode annotation

significance_directional

one-sided filter on effect column

Method has_pvalue()

Does the backend produce a usable p-value column?

Usage

ContrastConfiguration$has_pvalue()

Method clone()

The objects of this class are cloneable with this method.

Usage

ContrastConfiguration$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

cfg <- ContrastConfiguration$new(subject_id = "protein_Id")
cfg$effect_col
#> [1] "diff"
cfg$fdr_col
#> [1] "FDR"
cfg$has_pvalue()
#> [1] TRUE
cfg$supports_dea_qc
#> [1] TRUE

# SAINT-flavoured config
saint <- ContrastConfiguration$new(
  subject_id = "protein_Id",
  contrast_col = "Bait",
  effect_col = "log2_EFCs",
  score_col = "SaintScore",
  pvalue_col = NA_character_,
  fdr_col = "BFDR",
  supports_dea_qc = FALSE,
  needs_saint_annotation = TRUE
)
saint$has_pvalue()
#> [1] FALSE