Sequence: compute record lengths

Operation

Produce a tabular (id, length) table from a FASTA — one row per record — so the length can be filtered, sorted, joined, or thresholded with ordinary tabular tools (galaxy-tabular-patterns). The corpus uses:

toolshed.g2.bx.psu.edu/repos/devteam/fasta_compute_length (“Compute sequence length”).

This is a sequence → tabular move, narrower than sequence-fasta-tabular-interconvert: it drops the sequence and keeps only id + length. Keep its scope to the length table; aggregate assembly statistics (N50, total bases, GC) are a different concern handled by domain tools (fasta_stats, gfastats) that this pattern deliberately excludes — see iwc-sequence-operations-survey §2. Parameter names below are corpus-inferred from tool_state.

When to reach for it

• You need to filter or threshold records by length downstream (drop short contigs, keep scaffolds above N bp) and want the length as a column to filter on.
• You need a length distribution table for reporting or to join against other per-record data on the id.

For a one-shot length filter where you never need the table, a dedicated filter-by-length tool is more direct; this pattern is for when the length table itself is useful.

Parameters

• ref.ref_source: history — the FASTA comes from the history (connected input).
• keep_first: limit to the first N records; "0" (all) in the corpus.
• keep_first_word: use only the first whitespace-delimited header token as the id; false in the corpus (full header as id).

Idiomatic shape

tool_id: toolshed.g2.bx.psu.edu/repos/devteam/fasta_compute_length/1.0.3
tool_state:
  ref:
    ref_source: history
    input: { __class__: ConnectedValue }
    keep_first: "0"
    keep_first_word: false

Pitfalls

• The output is a table, not FASTA. It carries no sequence — you cannot reconstruct records from it. To keep the sequence alongside, use sequence-fasta-tabular-interconvert instead.
• keep_first_word changes the join key. With false the whole header is the id; a downstream join on a short accession will miss unless you also trim the header (or set keep_first_word: true).
• Not assembly statistics. This computes per-record length only. N50 / total-length / GC summaries are out of scope; reaching for gfastats/fasta_stats pulls in assembly-domain tooling this pattern intentionally avoids.

See also

• sequence-fasta-tabular-interconvert — the broader FASTA↔tabular move that keeps the sequence.
• galaxy-tabular-patterns — filter/sort/threshold the length table.
• galaxy-sequence-patterns — the sequence pattern map.
• iwc-sequence-operations-survey — corpus evidence.