Non-canonical ORFs GENCODE Phase II ncORFs Compr Track Settings
Assembly: Human Dec. 2013 (GRCh38/hg38) Data last updated at UCSC: 2026-05-19 04:39:19
Description
The three GENCODE ncORF tracks in the non-canonical ORF track container show
non-canonical translated open reading frames (ncORFs) identified
from ribosome profiling (Ribo-seq) data and mapped to the GENCODE annotation by the
GENCODE / TransCODE consortium,
as three tracks:
The Phase I catalog contains 7,264 unique human ncORFs called from Ribo-seq data
across seven publications and mapped to GENCODE v35. Only translations of 16 codons or above
and initiating from ATG start codons were incorporated. Redundant sense-overlapping ORFs were
merged. Of these, 3,085 ORFs were found by more than one publication, providing independent
replication evidence. This catalog was developed as part of an effort to standardize the
annotation of translated ORFs across reference databases including Ensembl/GENCODE, HGNC,
UniProtKB, and PeptideAtlas.
Phase II
The Phase II catalog nearly quadruples the Phase I set, defining 28,359 ncORFs in the
Comprehensive set, mapped to GENCODE v45. Compared to Phase I, additional published
Ribo-seq datasets were incorporated and the 16-codon minimum length and ATG-only start
codon restrictions from Phase I were removed.
Two subsets are provided for the Phase II data:
Comprehensive (28,359 ncORFs) is all mapped translations from the expanded catalog.
Primary (10,127 ncORFs) is a high-confidence subset selected using
translation signature scores derived from Ribo-seq data. These ncORFs demonstrate
translation evidence comparable to canonical protein-coding genes.
Display Conventions and Configuration
The default color scheme and available filter controls differ by track.
Phase I
Kozak strength colors (default)
The Phase I track colors items by Kozak strength by default. The
Color by dropdown on the track controls page offers two more options:
Evidence type and HLA class (HLA = Human Leukocyte Antigen; see below).
Kozak strength is based on two specific positions in the genomic sequence, counted
relative to the A of the ATG start codon: position −3 (three bases before the A)
and position +4 (one base after the G).
Golden amber: Strong match to the Kozak consensus. Both position −3 (A/G)
and position +4 (G) match.
Cornflower blue: Moderate match to the Kozak consensus. Only one of position
−3 (A/G) or position +4 (G) matches.
Gray: Weak match to the Kozak consensus. Neither position −3 (A/G)
nor position +4 (G) matches.
Black: Non-ATG start codon. The Kozak rule does not apply.
Light gray: Genomic context could not be read (for example, an ORF too
close to the edge of a chromosome).
Evidence type colors (alternative)
Select Color by: Evidence type to color items by their peptide evidence
from Deutsch et al. (see References).
Gold: Annotated as a peptidein by Deutsch et al. (121 ORFs, Table S12).
Steel blue: Detected only by HLA immunopeptidomics (1,667 ORFs).
Forest green: Detected only by non-HLA (whole-cell tryptic) proteomics
(51 ORFs).
Orange: Detected by both methods (21 ORFs).
Gray: No peptide evidence (5,404 ORFs). This includes ORFs whose only
non-HLA peptides also map to known proteins and were excluded by the authors.
HLA class colors (alternative)
Select Color by: HLA class to color items by whether their peptides
were detected in HLA Class I, Class II, or both:
Steel blue: Class I only (1,632 ORFs).
Crimson: Class II only (10 ORFs).
Orange: Both class I and class II (143 ORFs).
Gray: No HLA data (5,479 ORFs).
The Phase I track can be filtered by (in the order shown on the track controls page):
Kozak translational efficiency
HLA class
HLA evidence tier
HLA HPP guideline category
peptidein flag
Kozak strength
replicated status
Ribo-seq support
start codon
Mouseover for Phase I shows ORF name, host gene, Kozak strength and TE,
replicated status, peptidein flag, HLA evidence tier, and HLA peptide count.
Phase II
The Phase II Primary and Comprehensive tracks color items by Kozak strength using the
same scheme as Phase I. Peptide evidence fields are not included in the Phase II tracks,
so they can only be filtered by:
Kozak translational efficiency
Kozak strength
start codon
Mouseover for Phase II shows ORF name, host gene, start codon, and Kozak strength and TE.
Data Access
The raw data can be explored interactively with the
Table Browser or the
Data Integrator. The data can be accessed from
scripts through our API. The track names are
"gencNcOrfs" (Phase I), "gencNcOrfsPrimary" (Phase II Primary),
and "gencNcOrfsComprehensive" (Phase II Comprehensive).
For automated download and analysis, the genome annotations are stored in bigBed files that
can be downloaded from
our download server.
Individual regions or the whole genome annotation can be obtained using our tool
bigBedToBed, which can be compiled from the source code or downloaded as a precompiled
binary for your system. Instructions for downloading source code and binaries can be found
here.
Methods
Phase I Ribo-seq catalog
Mudge et al. (2022, see References) consolidated translation evidence from seven published
ribosome profiling datasets that used harringtonine or lactimidomycin treatment to enrich
for translation initiation sites. Ribo-seq reads were mapped to the GENCODE v35 annotation
on hg38. Only ATG-initiated ORFs of at least 16 codons were retained, and redundant
sense-overlapping ORFs were merged by taking the longest representative. This yielded 7,264
ncORFs across seven biotype classes: upstream ORFs (uORFs), upstream overlapping ORFs
(uoORFs), downstream ORFs (dORFs), downstream overlapping ORFs (doORFs), intronic ORFs
(intORFs), lncRNA-embedded ORFs, and ORFs on processed transcripts.
The catalog was developed as part of a reference-database coordination effort involving
Ensembl/GENCODE, HGNC, UniProtKB, HUPO/HPP, and PeptideAtlas.
Phase II Ribo-seq catalog
Chothani et al. (2025, see References) expanded the catalog by incorporating additional
Ribo-seq datasets across more cell types and tissues and mapping to GENCODE v45. The
ATG-start codon and 16-codon length restrictions were lifted to capture near-cognate
initiations and micropeptides. A data-driven scoring framework using ribosome occupancy
uniformity and P-site in-frame fraction identified a Primary subset of 10,127 ncORFs with
translation signatures comparable to canonical coding genes. The Comprehensive set contains
all 28,359 mapped ORFs.
Kozak strength and translational efficiency
Each ORF was annotated with its Kozak consensus strength by fetching the 11-base genomic
context around the start codon from hg38.2bit and classifying positions −3 and +4
relative to the A of the start codon. Both positions matching (A/G at −3 and G at +4)
gives Strong, one position matching gives Moderate, neither matching gives Weak, and a
non-ATG start gives non-ATG. A numeric translational efficiency (TE) score was also assigned
by looking up the 11-base context in the Noderer 2014 TE table (Mol Syst Biol 10:748,
PMID 25170020).
Peptide evidence and peptideins
Deutsch et al. (2026, see References) queried the 7,264 Phase I ORFs against two independent
PeptideAtlas mass spectrometry repositories. The HLA immunopeptidomics build was constructed
from HLA-I and HLA-II peptidomes across more than 100 HLA-typed donors spanning multiple
tissue types and cancer cell lines. Peptides were enriched by affinity purification and
identified by tandem mass spectrometry. The whole-cell tryptic proteomics build used
conventional shotgun proteomics from a broad range of cell lines and tissues. Spectra
were manually reviewed and classified according to the Prensner et al. tier system (Tier 1B =
numerous HPP-quality HLA peptides; Tier 2B = a single qualifying HLA peptide; Tier 1A/2A =
additional non-HLA evidence). The study introduced the term peptidein for a
translation product detectable by mass spectrometry but not yet annotatable as a protein due
to absent functional evidence. The authors reviewed 137 of the 7,264 Phase I ORFs for
annotation and classified 121 of them as peptideins (Table S12). A further 1,739 have HLA
or tryptic peptide evidence but were not classified as peptideins.
The supplementary data tables (Tables S2, S3, S6, S7, and S12) from Deutsch et al.
(2026, see References) were downloaded from the paper's supplementary materials.
Each table was joined to the Phase I bigGenePred by the short ORF identifier (e.g.,
c14riboseqorf80) using the script
addPeptideEvidence.py.
The script appended 14 new fields to all 7,264 Phase I items. The 5,353 ORFs that are not
in any of these tables receive default empty values so they remain visible in the track and
filterable on isPeptidein and related fields. Non-HLA peptides that map to known
proteins or are too short to be informative were excluded (Table S2, exclude
column), and the non-HLA peptide count includes only the remaining peptides.
The complete build procedure is documented in
ncOrfs.txt.
Credits
Thanks to Jonathan Mudge, Jorge Ruiz-Orera, John Prensner, Sebastiaan van Heesch, and the
GENCODE / TransCODE consortium for creating and maintaining these annotations.