ONGA
Ontology for Genomic Annotations
11 vocabularies + 5 schemas for describing genomic data files
Vocabularies
Closed value sets — the permissible values you choose from ("which one?").
DataType
162 terms
Algorithmic interpretation values — the data structure or processing result a file contains.
Describes HOW data was produced: peaks, signal tracks, alignments, count matrices.
FeatureType
75 terms
Biological interpretation values — the biological entity or concept the data represents.
Describes WHAT biology it captures: enhancers, TF binding sites, methylation states.
Format
12 terms
Genomic track file formats — EDAM-aligned.
BED, bigWig, narrowPeak, GFF3, VCF, BAM… mapped to EDAM.
Facet vocabularies
StrandOrientation
4 termsStrand sense (plus/minus/unstranded/bidirectional), factored out of compound DataType terms.
ReadMultiplicity
2 termsRead mapping-multiplicity (all reads vs. unique reads), factored out of compound DataType terms.
FilterStatus
2 termsQC filtering status (filtered vs. unfiltered), factored out of compound DataType terms.
Normalization
3 termsScaling normalization (raw vs. depth- vs. percentage-normalized), factored out of compound DataType terms.
Thresholding
3 termsRetention cut applied (IDR / mapping-quality / significance), factored out of compound DataType terms.
Derivation
2 termsEpistemic origin (observed vs. predicted), factored out of compound DataType / FeatureType terms.
ReferenceBuildSex
2 termsSex of the reference assembly build (male / female), factored out of compound DataType terms.
HaplotypeResolution
3 termsHaplotype resolution (allele-specific / haplotype-specific / phased), a content facet factored out of compound DataType / FeatureType terms.
Schemas
Descriptor properties you assert about a track ("describe this track"); slots draw on the vocabularies.
Track Format
3 properties
How the data is encoded — the file format and BED column conventions (FGA #1).
Its file_format slot draws on the Format vocabulary.
Track Interpretation
4 properties
What the data means — an algorithmic output_type + a biological feature_type (FGA #2).
Draws on the DataType and FeatureType vocabularies, plus strand and haplotype_resolution content facet slots — what the data is.
Track Provenance
5 properties
What was done to the data — the processing/derivation operations applied.
Read selection (read_multiplicity), filtering (filter_status), and scaling (normalization); extensible to derivation and bias-correction.
Track Geometry
14 properties
Its structural shape — measured geometric properties of a track's features (FGA #3).
Gaps, lengths, strands, values, edges (Gundersen et al. 2011).
Reference Genome
2 properties
The reference assembly it is defined against — every annotation is relative to a reference build.
An assembly identifier plus a
build_sex facet slot — the sex of the build, not the sample.
How they relate
The three core vocabularies are closed value sets, complemented by eight small
facet vocabularies. The five schemas are sets of
properties you assert about a track, with slots whose values come from those vocabularies or from primitives.
Track Format describes how a track is encoded (file format, BED columns).
Track Interpretation describes what it is / means, split into an algorithmic
output_type (drawing on DataType) and a biological
feature_type (drawing on FeatureType).
Track Provenance describes what was done to it — the processing/derivation operations
applied (read selection, QC filtering, scaling normalization, thresholding, and observed-vs-predicted derivation).
Track Geometry describes its structural shape (point vs interval vs signal vs graph) —
measured from the bytes, orthogonal to the others.
Reference Genome describes the reference assembly a track is defined against
(its identifier and build sex).
Properties that describe the sample or assay rather than the file's content —
anatomy, sample sex, method — are deliberately kept out of scope and delegated to external ontologies.