Skip to contents

Pairwise EIC similarity for xcms features

get_xcms_feature_EIC_similarity.Rd

Compute per-sample feature-by-feature EIC similarity matrices for pairs with |rtmed_i - rtmed_j| < rt_tol. Used by MSdev_group_feature_EIC. Results are stored as symmetric Matrix sparse matrices; absent entries are non-neighbor pairs (filled when densified for grouping; see absent_sim in xcms_group_feature_EIC). Pairwise scores are pooled across selected samples and split into n_chunks BiocParallel jobs (each chunk ships only the chromatogram columns it needs). EICs are expected to be zero-filled over their windows on the shared sample RT grid (via XChromatograms_subset_feature), so correlation reflects the full peak window rather than only the non-NA overlap.

Usage

get_xcms_feature_EIC_similarity(
  xcms.xcms,
  chroms,
  rt_tol = 5,
  expandRt = 2,
  min_width = 20,
  selected_sample = NULL,
  n_chunks = NULL,
  BPPARAM = BiocParallel::SerialParam()
)

Arguments

xcms.xcms

XCMSnExp / XcmsExperiment with feature definitions.

chroms

Chromatograms (e.g. from get_xcms_feature_chromatogram) with feature rownames matching featureDefinitions.

rt_tol

numeric(1). Maximum absolute RT difference (seconds) for which EIC similarity is computed. Default 5.

expandRt

numeric(1). Seconds added on each side of each feature's peakRtMin/peakRtMax window when cropping EICs via XChromatograms_subset_feature before correlation (requires peakRtMin/peakRtMax in chroms@featureDefinitions). Default 2.

min_width

numeric(1). Minimum RT window width (seconds) after expandRt; shorter windows are padded equally on both sides. Default 20.

selected_sample

NULL, integer index/indices, or sample name(s) (sample.name / chromatogram colnames). NULL uses all samples.

n_chunks

NULL or positive integer. Number of BiocParallel chunks into which n_samples * n_rt_pairs similarity jobs are split. When NULL (default), uses min(bpnworkers(BPPARAM), n_jobs) for backward compatibility. Set explicitly (e.g. 200L) on large batch projects to keep Snow payloads small; BPPARAM workers then control parallelism only.

BPPARAM

BiocParallel backend. Default SerialParam().

Value

Named list of feature-by-feature dgCMatrix similarity matrices (one per selected sample).