GenomicAnnotations.jl

Description

GenomicAnnotations is a package for reading, modifying, and writing genomic annotations in the GenBank and GFF3 file formats.

Installation

julia>]
pkg> add GenomicAnnotations

Usage

GenBank and GFF3 files are read with readgbk(input) and readgff(input), which return vectors of Records. input can be an IOStream or a file path. GZipped data is unzipped automatically if a filename ending in ".gz" is passed as input. If we're only interested in the first chromosome in example.gbk we only need to store the first element.

chr = readgbk("test/example.gbk")[1]

Records have five fields, name, header, genes, genedata, and sequence. The name is read from the header, which is stored as a string. The annotation data is stored in genedata, but generally you should use genes to access that data. For example, it can be used to iterate over annotations, and to modify them.

for gene in chr.genes
    gene.locus_tag = "$(chr.name)_$(gene.locus_tag)"
end

chr.genes[2].locus_tag = "test123"

The locus of a Gene can be retrieved with locus(gene), and updated with locus!(gene, newlocus). The easiest way to create a locus is to use the constructor Locus(s), which takes an AbstractString s and parses it as a GenBank locus string as defined here: https://www.insdc.org/submitting-standards/feature-table/#3.4. Note that remote entry descriptors have not been implemented.

# The following are all equivalent
locus!(gene, "complement(join(1..100,200..>300))")
locus!(gene, Locus("complement(join(1..100,200..>300))"))
locus!(gene, Complement(Join(ClosedSpan(1:100), OpenRightSpan(200:300))))

position(gene) can be used as shorthand for locus(gene).position to retrieve the chromosomal positions included in the locus, excluding all metadata such as strandedness. The return type depends on the locus type, but is quaranteed to iterate over the individual positions.

for i in position(gene)
    print(parent(gene).sequence[i])
end
# is equivalent to
print(sequence(gene))

Accessing properties that haven't been stored will return missing. For this reason, it often makes more sense to use get() than to access the property directly.

# chr.genes[2].pseudo returns missing, so this will throw an error
if chr.genes[2].pseudo
    println("Gene 2 is a pseudogene")
end

# ... but this works:
if get(chr.genes[2], :pseudo, false)
    println("Gene 2 is a pseudogene")
end

The macro @genes can be used to filter through the annotations. The macro takes a Record or a Vector{Record}, followed by any number of expressions that will be evaluated for each gene. The keyword gene is used to refer to the individual Genes. @genes can also be used to modify annotations. Gene attributes can be referred to using Symbols.

@genes(chr, feature(gene) == "CDS")  # Returns all coding regions
@genes(chr, length(gene) > 300) # Returns all features longer than 300 nt
@genes(chr, iscomplement(gene)) # Returns all features on the complement strand
@genes(chr, ismissing(:product)) # Returns all features for which the attribute "product" has not been set

# Some short-hand forms are available to make life easier:
#     `CDS` expands to `feature(gene) == "CDS"`, and
#     `get(s::Symbol, default)` expands to `get(gene, s, default)`
# The following two are thus equivalent:
@genes(chr, feature(gene) == "CDS", occursin("glycoprotein", get(gene, :product, "")))
@genes(chr,                   CDS,  occursin("glycoprotein", get(      :product, "")))

# All arguments have to evaluate to `true` for a gene to be included, so the following expressions are equivalent:
@genes(chr, feature(gene) == "CDS", length(gene) > 300)
@genes(chr, (feature(gene) == "CDS") && (length(gene) > 300))

# `@genes` returns a `Vector{Gene}`. Attributes can be accessed with dot-syntax, and can be assigned to:
@genes(chr, :locus_tag == "tag03")[1].pseudo = true
@genes(chr, CDS, ismissing(:gene)).gene .= "unknown"

Gene sequences can be accessed with sequence(gene). For example, the following code will write the translated sequences of all complete protein-coding genes to a file:

using BioSequences
using FASTX
open(FASTA.Writer, "proteins.fasta") do w
    for gene in @genes(chr, CDS, iscomplete(gene))
        aaseq = GenomicAnnotations.sequence(gene; translate = true)
        write(w, FASTA.Record(gene.locus_tag, get(:product, ""), aaseq))
    end
end

Genes can be added using addgene!, and sort! can be used to make sure that the resulting annotations are in the correct order for printing. delete! is used to remove genes.

newgene = addgene!(chr, "regulatory", 670:677)
newgene.locus_tag = "reg02"
sort!(chr.genes)

# Genes can be deleted. This works for all genes where `:pseudo` is `true`, and ignores genes where it is `false` or `missing`
delete!(@genes(chr, :pseudo))
# Delete all genes 60 nt or shorter
delete!(@genes(chr, length(gene) <= 60))

Individual genes, and Vector{Gene}s are printed in GBK format. To include the GBK header and the nucleotide sequence, printgbk(io, chr) can be used to write them to a file. printgff(io, chr) prints the annotations as GFF3, in which case the GenBank header is lost.

println(chr.genes[1])
println(@genes(chr, CDS))

open(GenBank.Writer, "updated.gbk") do w
    write(w, chr)
end