cdskit longestorf

cdskit longestorf searches six translated frames (3 on + strand and 3 on - strand) and returns one ranked ORF candidate for each input sequence. The default prioritizes complete candidates before length.

Example

Command

cdskit longestorf --seq_file unaligned_nt.fasta --out_file longest_orf.fasta --codon_table 1 --annotate_seq_name yes

unaligned_nt.fasta

>seq1
AAAATGAAACCCTAGGGGATGAAAAAAACCCCTGAATGATGCCCTAA

longest_orf.fasta

>seq1 strand=+ frame=1 start=19 end=39 nt_len=21 aa_len=7 category=complete
ATGAAAAAAACCCCTGAATGA

Multiple ORFs in one sequence

For seq1, multiple complete ORFs exist in the same frame (+, frame 1). cdskit longestorf selects the longest one.

Sequence (+, frame 1 codons):
AAA | ATG AAA CCC TAG | GGG | ATG AAA AAA ACC CCT GAA TGA | TGC CCT ...
      [ORF1: 12 nt]              [ORF2: 21 nt, selected]

Candidate priority

Candidates are prioritized in the following order:

1. complete: start codon + in-frame stop codon
2. partial: start codon to frame end (no in-frame stop)
3. no_start: longest stop-free segment (when no start-based CDS is found)

Notes

• longestcds is a deprecated alias with the same behavior; use longestorf in new commands.
• Input sequences do not need to be aligned.
• Output sequence orientation follows the predicted coding strand.
• Coordinates in output description (start, end) are reported on the original input strand coordinates (1-based), with strand indicated by strand=+/-.
• Header annotation (strand=... frame=... start=...) is optional via --annotate_seq_name yes|no (default: no).
• aa_len is nt_len / 3; for a complete candidate it includes the terminal stop position. It is not the length of the protein after removing *.

Explicit selection and candidate reports

The default is --selection complete-first: category priority above, then length. --selection longest instead compares length first across all start-based candidates and maximal stop-free segments in all six frames, then category. A no_start candidate may win even when start-based candidates exist. The search is not limited to one longest complete candidate per strand.

Both modes resolve remaining ties deterministically: plus strand, then the existing stop/start flags, smaller frame number, and smaller original start coordinate. Length is nucleotide length, including the terminating codon for a complete candidate. This convention is shared by both modes.

cdskit longestorf --seq_file transcripts.fasta --out_file candidates.fasta \
  --selection longest --report candidates.json --annotate_seq_name yes

--report records all start-to-first-definite-stop (or frame-end) candidates and maximal stop-free segments, including unselected ones, with ranks, ties before deterministic ordering, coordinates, stop evidence and uncertainty. Nested starts are included; arbitrary subsegments are not enumerated. Only the selected candidate contains output_seq; unselected sequences are reconstructed from coordinates and strand. Counts are computed from per-frame prefix sums, so nested candidates do not repeatedly scan or copy the same long sequence. Reports still contain one row per candidate. sort_key records the actual selection policy, and missing/ambiguous counts supplement stop uncertainty. Coordinates refer to the original input; output sequence is in coding orientation.

Codes 27/28/31 do not terminate a candidate at a dual-coding codon based solely on table membership. Such boundaries are unconfirmed, and the candidate may extend to frame end. An explicit annotated CDS can instead be checked with translate --complete_cds yes. Standard-code TAR is a definite boundary; TAN is not. See codon semantics.

Neither selection mode identifies a functional gene. For example, the default selects the 6 nt complete ATGTAA even if a much longer partial region follows. Longer candidates are not inherently more accurate gene annotations; compare coordinates to independent annotation, homology and other biological evidence.

Citation

No separate method paper is designated for this utility. Report the CDSKIT version, command and options; see citing CDSKIT.