Interval: window or flank features

Operation

Extend each feature by a fixed (or fractional) amount to build a neighborhood window around it — summit ± N bp, gene + upstream promoter, crosslink site ± N. toolshed.g2.bx.psu.edu/repos/iuc/bedtools/bedtools_slopbed (“bedtools SlopBed”) does this, clamping the result to chromosome bounds via a genome file so windows never run off the ends.

Single corpus occurrence (atacseq), but it recurs in the GTN training corpus — the transcriptomics/clipseq tutorial uses SlopBed to widen crosslink sites before fetching sequence, the same “extend to a neighborhood” move in a different domain. That second, independent use is why it earns a page rather than living only as a recipe ingredient.

Parameter names are corpus-inferred from tool_state.

When to reach for it

• Build fixed windows around point/summit features before quantifying coverage (the first step of interval-windowed-coverage).
• Add flanks to features (promoter regions upstream of genes; padding around peaks) before an overlap or extract step.
• Symmetric (b) or asymmetric (l/r) extension.

To collapse the windows you build (so overlapping ones aren’t double-counted), follow with interval-merge-overlapping.

Parameters

• inputA: connected interval set.
• genome_file_opts.genome_file_opts_selector: loc (a built-in genome) or hist (a history chromosome-sizes file). Required — slop needs chromosome lengths to clamp extensions; without it, windows can extend past chromosome ends into invalid coordinates.
• addition.addition_select: b (both sides, the corpus value), l (left/upstream only), or r (right/downstream only).
  • addition.b (or .l/.r): the extension amount. "500" in atacseq — 500 bp each side.
• pct: false (corpus) treats the amount as base pairs; true treats it as a fraction of each feature’s length.
• strand: false (corpus) ignores strand; true makes l/r mean upstream/downstream relative to strand.
• header: false in the corpus.

tool_id: toolshed.g2.bx.psu.edu/repos/iuc/bedtools/bedtools_slopbed/2.31.1+galaxy0
tool_state:
  inputA: { __class__: ConnectedValue }
  genome_file_opts:
    genome_file_opts_selector: loc
    genome: { __class__: ConnectedValue }
  addition: { addition_select: b, b: "500" }
  pct: false
  strand: false
  header: false

Pitfalls

• Label drift: bp vs kb. The atacseq step is labelled “get summits +/-500kb” and renamed “1kb around each summit”, but b: "500" is 500 bp — a 1 kb total window, not 500 kb. Loose labels are common in IWC; trust the parameter, not the step name.
• The genome file is mandatory and load-bearing. It is the only thing keeping extensions inside valid coordinates. A missing or mismatched genome file (wrong assembly) produces silently wrong windows that overrun chromosome ends.
• pct: true changes units. With pct: true, b: "0.5" means “extend by half the feature length each side,” not 0.5 bp. Easy to conflate with the base-pair mode.
• strand: false makes l/r genomic, not biological. For strand-aware upstream/downstream (promoters), set strand: true, or l/r will be wrong for minus-strand features.

See also

• galaxy-interval-patterns — the interval pattern map.
• interval-windowed-coverage — the slop → merge → coverage recipe this op opens.
• interval-merge-overlapping — collapse overlapping windows after slop.
• iwc-interval-operations-survey — corpus evidence (and the GTN second-example note).