Bio Ribo Seq Orf Detection

Version Compatibility

Reference examples tested with: RiboCode 1.2+, ORFquant 1.0+, ORFik 1.22+, DESeq2 1.42+, pandas 2.2+

Before using code patterns, verify installed versions match. If versions differ:

• Python: pip show <package> then help(module.function) to check signatures
• R: packageVersion('<pkg>') then ?function_name to verify parameters
• CLI: <tool> --version then <tool> --help to confirm flags

If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.

ORF Detection

"Detect translated ORFs from my Ribo-seq data" -> Identify actively translated open reading frames (uORFs, internal ORFs, dORFs, novel ORFs) using 3-nucleotide periodicity (not mere coverage) as the evidence of translation, then classify and quantify them.

• CLI: RiboCode for periodicity-based de novo ORF calling
• R: ORFquant for isoform-aware quantification; ORFik as the general toolkit

The discriminating signal is PERIODICITY: a translated ORF shows footprint P-sites in frame 0 (F0 >> F1, F2). Coverage alone is not evidence of translation. Every method needs a correct per-read-length P-site offset first (see ribosome-periodicity).

ORF-type nomenclature (Mudge 2022 standard)

The GENCODE-led standard (Mudge et al 2022) defines the umbrella term "Ribo-seq ORF" and six positional categories. These are positional, not modification-based (N-terminal extensions are not part of the scheme).

Related terms: sORF/smORF (<100 codons, product = microprotein), annotated CDS, novel. Catalogs: sORFs.org, OpenProt.

Near-cognate start codons (the ATG-only blind spot)

Initiation occurs at AUG and at near-cognate codons differing from AUG by one base; the biologically used set is CUG, GUG, ACG, UUG, AUU, AUC, AUA (AAG/AGG also differ by one base but initiate negligibly). CUG is the dominant near-cognate start (~16% of mapped initiation sites; AUG remains >50%). uORFs ESPECIALLY use near-cognate starts, so an ATG-only scanner misses the majority of real uORFs. Periodicity-based callers can be configured with alternative starts (RiboCode -A); the manual finder below is ATG-only and is a teaching toy unless extended.

Tool selection

ORFik and ORFquant are DISTINCT packages (different authors, repos, methods): ORFik (Tjeldnes 2021, general toolkit) is not the same as ORFquant (Calviello 2020, dedicated isoform-aware caller). Do not install ORFik expecting ORFquant.

Call ORFs de novo with RiboCode

Goal: Identify periodicity-significant ORFs, including uORFs at near-cognate starts.

Approach: Prepare transcript annotation, run metaplots to select periodic read lengths and their P-site offsets, then run RiboCode with optional alternative starts.

# Step 1: annotation
prepare_transcripts -g annotation.gtf -f genome.fa -o ribocode_annot

# Step 2: metaplots picks periodic read lengths + per-length P-site offsets -> config .txt
#   (read lengths come from THIS step, NOT from a -l flag)
metaplots -a ribocode_annot -r transcriptome.bam -o metaplots_out

# Step 3: call ORFs. -A adds near-cognate starts; -l is the longest-ORF toggle (yes/no)
RiboCode -a ribocode_annot -c metaplots_out_pre_config.txt \
    -A CTG,GTG -l no -p 0.05 -o ribocode_result

RiboCode works in transcript coordinates, so the -r input is the TRANSCRIPTOME-projected BAM (Aligned.toTranscriptome.out.bam), not the genome BAM; a genome BAM silently misbehaves. Its core test is a MODIFIED WILCOXON SIGNED-RANK test on the per-codon P-site frame distribution (a separate binomial file is a secondary output). The -l flag toggles longest-ORF selection; it is NOT a read-length list.

Parse and classify RiboCode output

Goal: Split called ORFs by the standard categories.

Approach: Read the tabular result and group on the ORF_type column, whose RiboCode values are annotated, uORF, dORF, Overlap_uORF, Overlap_dORF, Internal, novel.

import pandas as pd

def load_ribocode_orfs(path):
    '''Load the RiboCode result table (<output_name>.txt) and group by ORF_type.'''
    df = pd.read_csv(path, sep='\t')
    groups = {t: df[df['ORF_type'] == t] for t in df['ORF_type'].unique()}
    return df, groups

RiboCode writes the result as <output_name>.txt (plus a <output_name>_collapsed.txt), e.g. ribocode_result.txt for -o ribocode_result; its columns include ORF_ID, ORF_type, transcript/genome start-stop, pval_combined, and adjusted_pval.

Quantify ORFs isoform-aware with ORFquant

Goal: Quantify ORF-level translation while resolving footprints across overlapping isoforms.

Approach: Prepare annotation once, feed Ribo-seQC-prepared input, then run the master function.

library(ORFquant)

prepare_annotation_files(annotation_directory = "annot/",
                         twobit_file = "genome.2bit",
                         gtf_file = "annotation.gtf")
# Ribo-seQC writes a for_ORFquant object from the Ribo-seq BAM; pass it here
run_ORFquant(for_ORFquant_file = "sample_for_ORFquant",
             annotation_file = "annot/annotation.gtf_Rannot",
             n_cores = 4)

For uORF discovery in a general R workflow, ORFik provides findUORFs() and the true P-site shift is detectRibosomeShifts() then shiftFootprints() (there are no p_offsets/lengths arguments on fimport).

Manual ORF scan (teaching reference, ATG-only)

Goal: Illustrate ORF finding mechanics; not a substitute for a periodicity caller.

Approach: Scan three frames for start-to-stop pairs. This finds only ATG starts and uses coverage, not periodicity, so it misses near-cognate uORFs and cannot confirm active translation on its own.

def find_orfs(seq, min_codons=10):
    '''Find ATG-to-stop ORFs in all three frames (ATG-only: a teaching toy).'''
    seq = str(seq).upper()
    stops = {'TAA', 'TAG', 'TGA'}
    orfs = []
    for frame in range(3):
        i = frame
        while i < len(seq) - 2:
            if seq[i:i+3] == 'ATG':
                for j in range(i + 3, len(seq) - 2, 3):
                    if seq[j:j+3] in stops:
                        if (j + 3 - i) >= min_codons * 3:
                            orfs.append({'start': i, 'end': j + 3, 'frame': frame})
                        i = j
                        break
            i += 3
    return orfs