Sequence: extract or mask by region

Operation

Turn coordinates into sequence: given regions (a BED of intervals, or a GFF of annotated features) and a genome/source FASTA, produce sequence records — either the sub-sequences at those regions, or the source with those regions masked. Three corpus moves share this shape; they differ in what the region set is and whether the output is the extract or the masked whole.

This page sits on the interval/annotation ↔ sequence bridge. The regions are produced by interval algebra (galaxy-interval-patterns) or annotation tools; the output is sequence, which is why these operations live in the sequence MOC and were held out of the interval MOC (see iwc-sequence-operations-survey §5 and the interval survey’s scope note). Parameter names below are corpus-inferred from tool_state.

Move 1 — extract FASTA at intervals (bedtools getfasta)

toolshed.g2.bx.psu.edu/repos/iuc/bedtools/bedtools_getfastabed (“bedtools getfasta”). Pull the sub-sequence at each BED interval out of a FASTA.

• fasta_source.fasta_source_selector: history in the corpus — the source FASTA is a history dataset (connected); cached would use a built-in genome.
• input: the BED of intervals (connected).
• nameOnly / split / strand / tab: output toggles, all false in the corpus (default name handling, no strand-aware reverse-complement, FASTA not tabular output).

tool_id: toolshed.g2.bx.psu.edu/repos/iuc/bedtools/bedtools_getfastabed/2.30.0+galaxy1
tool_state:
  fasta_source: { fasta_source_selector: history, fasta: { __class__: ConnectedValue } }
  input: { __class__: ConnectedValue }
  nameOnly: false
  split: false
  strand: false
  tab: false

Move 2 — mask FASTA regions by BED (bedtools maskfasta)

toolshed.g2.bx.psu.edu/repos/iuc/bedtools/bedtools_maskfastabed (“bedtools maskfasta”). Keep the whole FASTA but replace the bases inside the BED regions. The corpus instances (mgnify ITS and complete pipelines) rename the masked output “ITS FASTA files”.

• fasta / input: source FASTA and the BED of regions to mask (connected).
• soft: false = hard mask (replace bases with the mask character); true would lowercase instead.
• mc: mask character, N in the corpus.
• fullheader: false (use the short header).

tool_id: toolshed.g2.bx.psu.edu/repos/iuc/bedtools/bedtools_maskfastabed/2.31.1
tool_state:
  fasta: { __class__: ConnectedValue }
  input: { __class__: ConnectedValue }
  soft: false
  mc: N
  fullheader: false

Move 3 — extract transcript/CDS FASTA from annotation (gffread)

toolshed.g2.bx.psu.edu/repos/devteam/gffread (“gffread”). Given an annotation GFF/GTF and the genome FASTA, emit the spliced transcript or CDS sequences. Every corpus use is inside a genome_annotation pipeline (maker, braker3, helixer, lncRNAs) — the operation is sequence extraction, but its region set comes from an annotation step rather than interval algebra.

• input: the GFF/GTF (connected).
• reference_genome.genome_fasta: the genome FASTA the features index into (connected).
• The corpus leaves filtering/merging/attribute options at defaults; reach for them only when the annotation needs cleaning first.

When to reach for each

• Region → just the sub-sequences: getfasta.
• Region → the whole source with those spans hidden: maskfasta (e.g. masking low-confidence or repeat regions before a consensus or search).
• Annotation features → their spliced sequences: gffread (transcripts/CDS from a gene model).

Pitfalls

• Strand matters for extraction. getfasta with strand: false returns the plus-strand sequence regardless of the feature’s strand; set strand: true when you need strand-aware (reverse-complemented) extraction. The corpus extracts plus-strand.
• Soft vs hard mask is a downstream contract. maskfasta soft: false writes Ns (most tools treat as ambiguous); soft: true lowercases (repeat-masker convention some aligners honor). Picking the wrong one silently changes how a downstream tool reads the masked bases.
• The BED must match the FASTA’s coordinate system and chrom names. A region whose chrom does not match a FASTA header yields nothing extracted / nothing masked — no error, just empty or unchanged output. The region set usually comes from galaxy-interval-patterns; keep names consistent.
• gffread is annotation-domain-resident. It is a sequence-extraction operation, but it presumes a cleaned gene model; if your features come from interval algebra rather than an annotator, getfasta is the simpler fit.

See also

• galaxy-interval-patterns — produces the BED regions feeding moves 1–2; cross-MOC bridge.
• galaxy-sequence-patterns — the sequence pattern map.
• sequence-fasta-tabular-interconvert — when the extracted records then need header edits.
• iwc-sequence-operations-survey — corpus evidence and the bridge scope note.