Tabular: synthesize BED from 3-column input

Tool

toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_awk_tool. This is one of the operation-named awk recipe pages from iwc-tabular-operations-survey §7.

When to reach for it

Use this when a three-column interval-like table needs to become BED-like six-column rows:

chrom_or_seqid<TAB>start<TAB>end

to:

chrom_or_seqid<TAB>start0<TAB>end<TAB>name<TAB>score<TAB>strand

The corpus-attested use is MGnify rRNA prediction: separate SSU/LSU forward/reverse interval outputs become hidden BED datasets, then forward and reverse files are concatenated per target. This is not a general tabular format-conversion page.

Parameters

• code: awk program. Corpus uses a quoted one-liner.
• infile: connected three-column input.
• Output metadata: corpus examples set change_datatype: bed and hide: true on the output.

Version pins differ (9.5+galaxy0 and 9.3+galaxy1) with stable parameter shape. Prefer the newest/current pin already used in the workflow.

Idiomatic shapes

Forward-strand BED:

tool_id: toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_awk_tool/9.5+galaxy0
tool_state:
  code: 'BEGIN {OFS="\t"} {print $1, $2 - 1, $3, "forward", "1", "+"}'
  infile: { __class__: ConnectedValue }
out:
  - id: outfile
    change_datatype: bed
    hide: true

Reverse-strand BED:

tool_id: toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_awk_tool/9.5+galaxy0
tool_state:
  code: 'BEGIN {OFS="\t"} {print $1, $2 - 1, $3, "reverse", "1", "-"}'
  infile: { __class__: ConnectedValue }
out:
  - id: outfile
    change_datatype: bed
    hide: true

Pitfalls

• Start coordinate off-by-one. BED start is 0-based; corpus subtracts 1 from $2.
• Do not subtract from end. BED end stays $3 in the corpus shape.
• Set OFS="\t". Without it, print $1, ... emits spaces.
• No header handling is shown. Corpus inputs are treated as data-only. If input has a header, add NR > 1 deliberately.
• BED datatype matters. Corpus sets change_datatype: bed; do not leave the output as generic tabular if downstream expects BED.
• Name and score are constants. This pattern does not derive BED name or score from input columns.
• Forward and reverse are separate steps. Do not emit both strands from one row unless downstream expects duplicate intervals in one file.

See also

• iwc-tabular-operations-survey — §2g and §7 awk split decision.
• tabular-prepend-header — another operation-named awk recipe.
• tabular-compute-new-column — simple per-row arithmetic when the output remains tabular, not BED.