The ORF and ORFCollection Types

GeneFinder uses two complementary types for representing Open Reading Frames:

• ORF: A lightweight, coordinate-based structure storing location and metadata

• ORFCollection: A container bundling ORFs with a view of their source sequence

This design ensures ORFs are composable, serializable, and independent of global state.

The ORF Type

Individual ORFs store only coordinates and metadata:

orf = collection[1]

leftposition(orf)   # Start position
rightposition(orf)  # End position
strand(orf)         # PSTRAND or NSTRAND
frame(orf)          # Reading frame (1, 2, or 3)
features(orf)       # NamedTuple of features
finder(orf)         # The method that found it

Validation Rules

The ORF constructor validates:

• Strand must be PSTRAND or NSTRAND

• Frame must be 1, 2, or 3

• Range length must be divisible by 3

• Range length must be ≥ 6 (start + stop codons)

• Features must be a NamedTuple

# This will error - frame must be 1, 2, or 3
ORF{NaiveFinder}(1:33, PSTRAND, Int8(4))
# ERROR: ArgumentError: Invalid frame: 4, expected 1, 2, or 3

This architecture provides:

• No global state: Source sequence is explicitly bundled

• Composability: Works in any function or module context

• Serialization: Collections can be saved/loaded

• Type safety: Compile-time guarantees on finder method

• Efficiency: Views used where possible, validation at construction

ORFCollection

All GeneFinderMethod implementations return an ORFCollection:

using BioSequences, GeneFinder

seq = dna"ATGATGCATGCATGCATGCTAGTAACTAGCTAGCTAGCTAGTAA"
collection = findorfs(seq)

# The collection bundles ORFs with a view of the source
typeof(collection)  # ORFCollection{NaiveFinder, LongSubSeq{DNAAlphabet{4}}}

# Source is always stored as a view (memory efficient)
typeof(source(collection))  # LongSubSeq{DNAAlphabet{4}}

# Access ORFs
orfvector(collection)      # Vector of ORFs
collection[1]         # First ORF
length(collection)    # Number of ORFs

Note

The source sequence is always stored as a LongSubSeq (view), which avoids copying the sequence data while maintaining a reference to the original. This is both memory-efficient and ensures the collection stays synchronized with the source.

Sequence Extraction

The ORFCollection provides clean sequence extraction using the sequence function:

# Extract by index
seq1 = sequence(collection, 1)

# Extract by ORF
orf = collection[1]
seq1 = sequence(collection, orf)

# Extract all sequences using broadcasting
all_seqs = sequence.(Ref(collection), collection.orfs)

Translation

Translate ORF sequences to amino acids using BioSequences.translate:

using BioSequences

# Translate a single ORF
protein = translate(collection, 1)

# Translate multiple ORFs with a comprehension
proteins = [translate(collection, i) for i in eachindex(collection)]