4.8.1 smORF scanner

Function

smorf_scanner scans transcript-centric candidate small open reading frames (smORFs) from a genome FASTA file and a genePred annotation file.

It reconstructs spliced transcript sequences from the genome, searches candidate ORFs using user-defined start codons and ORF length limits, classifies ORF categories by their position relative to the annotated CDS, optionally marks nested and overlapping ORFs, and writes four output files: an ORF annotation table, a full ORF information table, and the nucleotide and peptide FASTA files.

The scanner is designed for large genomes: each reading frame is translated exactly once and candidate ORFs reuse slices of the cached frame peptide, so nested ORFs are not translated independently. Multiprocessing (-t) splits the transcript set across workers.

Workflow

genome FASTA + genePred annotation → smorf_scanner → smorf.raw.genePred / message.txt / nt.fa / pep.fa → smorf_cluster

The command runs in two steps:

1. Checking the input arguments – validate the genome and annotation files and the ORF scanning parameters.
2. Scanning ORFs from transcript sequences – reconstruct transcript sequences, scan the reading frames, classify categories, and write all output files. With -t 1 a single-process streaming path is used; with -t > 1 transcripts are processed in parallel and the output is written in balanced partitions.

Input files

Input	Required	Description
Genome FASTA	Yes	Genome sequence used to reconstruct spliced transcript sequences.
genePred annotation	Yes	Transcript annotation in genePred format.

Parameters

Parameter	Required	Description
-g, --genome	Yes	Input genome sequence file in FASTA or FASTA.GZ format.
-a, --annotation	Yes	Input transcript annotation file in genePred format.
-o, --out-prefix	No	Prefix of all output files. Default: ORF.
-p, --orf-prefix	No	Prefix used to generate stable ORF identifiers. Default: ORF.
-s, --start-codons	No	Comma-separated start codons used for ORF scanning. RNA U is converted to T. Default: ATG.
-m, --min-aa	No	Minimum ORF peptide length in amino acids. Default: 8.
-M, --max-aa	No	Maximum ORF peptide length in amino acids. Default: 10000.
-x, --scan-strand	No	Strand mode for scanning: sense, antisense, or both. Default: sense.
-u, --kozak-up	No	Number of upstream nucleotides extracted for the Kozak context. Default: 6.
-d, --kozak-down	No	Number of downstream nucleotides after the start codon extracted for the Kozak context. Default: 6.
-I, --include-stop	No	Keep the stop-codon symbol in peptide sequences. Default: False.
-O, --mark-overlap	No	Mark nested or overlapping ORFs after scanning. Default: False.
-R, --remove-discarded	No	Remove same-frame internal ORFs that the classifier labels as discarded. Default: False.
-t, --thread	No	Number of worker processes. Default: 1.

Output files

Assuming -o gmx4, the command writes:

Output	Description
gmx4.genePred	ORF annotation in a genePredExt-like format.
gmx4.message.txt	Full ORF information table. This is the main input for smorf_cluster.
gmx4.nt.fa	Candidate ORF nucleotide sequences in FASTA format.
gmx4.pep.fa	Candidate ORF peptide sequences in FASTA format.

ORF information table (gmx4.message.txt)

One row per candidate ORF, tab-separated:

orf_id  gene_id  transcript_id  chrom  strand  source_strand  category  priority  overlap_type
frame  tx_orf_start  tx_orf_end  genomic_start  genomic_end  start_codon  stop_codon
nt_length  aa_length  kozak_seq  completeness  exon_count  exon_starts  exon_ends
kozak_start_index  ambiguous_codon_count

Column	Description
orf_id	Generated stable ORF identifier (for example ORF00000001).
gene_id / transcript_id	Source gene and transcript of the ORF.
chrom / strand	Genomic location and strand of the ORF.
source_strand	Scanning orientation relative to the transcript: sense or antisense.
category	Positional ORF category assigned by the classifier (see below).
priority	Overlap-filtering priority: primary, secondary, or discarded.
overlap_type	Relationship to overlapping ORFs (only filled with -O).
frame	Reading frame in the scanned sequence.
tx_orf_start / tx_orf_end	ORF boundaries in transcript coordinates.
genomic_start / genomic_end	Minimum/maximum genomic coordinate covered by the ORF.
start_codon / stop_codon	Detected start codon; stop codon, or NA for a partial ORF.
nt_length / aa_length	ORF nucleotide length (including the terminal stop codon when present) and peptide length.
kozak_seq	Fixed-width, N-padded start-codon context.
completeness	complete or 3prime_partial.
exon_count / exon_starts / exon_ends	ORF exon structure in ascending genomic order.
kozak_start_index	Zero-based start-codon index in kozak_seq.
ambiguous_codon_count	Number of ORF codons containing non-ACGT characters.

Category classification

ORFs are classified by their position relative to the annotated CDS of the transcript:

Category	Meaning
annotated_ORF	Matches the annotated CDS (highest priority).
uORF	Fully upstream of the CDS (5' UTR).
dORF	Fully downstream of the CDS (3' UTR).
emORF	Contains the CDS start or end (extended / merged ORF).
same_frame_iORF	Internal to the CDS, same reading frame (marked discarded by default).
iORF	Internal to the CDS, different reading frame.
overlap_uORF	Overlaps the CDS start.
overlap_dORF	Overlaps the CDS end.
lncORF	Found on a non-coding transcript.
antisense_ORF	Found on the antisense strand.
other_ORF	Any other configuration.

Sequence FASTA files

The FASTA headers carry the ORF metadata, for example:

>ORF00000001 gene=GlmaCp001 transcript=GlmaCp001 type=annotated_ORF strand=- length=1062
ATGACTGCAATTTTAGAGAGACGCGAGAGCGAAAGCCTATGGGGTCGCTTCTGTAACTGG

• gmx4.nt.fa – nucleotide sequences in coding orientation; the header includes length=.
• gmx4.pep.fa – translated peptide sequences; the header includes aa_length=. The terminal stop symbol is kept only with -I.

Examples

Use the common scanning setup on a large genome

cd ./sce/5.smorf/01.scanner

smorf_scanner \
  --genome ~/gmx/genome/GCF_000004515.6_Glycine_max_v4.0_genomic.fna \
  --annotation ~/gmx/norm/gmx4.genepred \
  --out-prefix gmx4 \
  --start-codons ATG,CTG,GTG,TTG,ACG,ATA,ATT,ATC \
  --min-aa 8 \
  --max-aa 10000 \
  --scan-strand sense \
  --kozak-up 6 \
  --kozak-down 6 \
  --mark-overlap \
  --remove-discarded \
  --thread 20 \
  &> gmx4.scan.log

This example scans the soybean genome (93,168 transcripts). With 20 workers it finished in about a minute and wrote 13,347,430 candidate ORFs into gmx4.genePred, gmx4.message.txt, gmx4.nt.fa, and gmx4.pep.fa (approximately 1.5–2.8 GB each).

Notes

• --scan-strand sense scans ORFs on the annotated transcript strand only; --scan-strand antisense scans the reverse-complemented transcript sequences; --scan-strand both covers both but substantially increases the number of candidates.
• -O/--mark-overlap is recommended when downstream filtering should distinguish primary, nested, or overlapping candidates; without it overlap_type is left at none.
• -R/--remove-discarded removes complete same-frame internal ORFs that the classifier labels as discarded; keep it enabled for cleaner candidate sets.
• The scanner only generates candidate ORFs: scanner output alone should not be treated as evidence of translation.
• Scanner output is a candidate set. Use smorf_cluster (4.8.2) and smorf_evidence (4.8.3) before prioritizing translated smORFs.