Changed Gene structure and API

BioJulia · Dec 10, 2019 · d70ea7f · d70ea7f
1 parent 5a19426
commit d70ea7f
Show file tree

Hide file tree

Showing 5 changed files with 139 additions and 120 deletions.
diff --git a/src/GenomicAnnotations.jl b/src/GenomicAnnotations.jl
@@ -7,6 +7,7 @@ using GZip
 
 export Chromosome, Gene, AbstractGene, GeneDataView, Locus
 export readgbk, sequence, iscomplement, addgene!, pushproperty!, printgbk
+export feature, index, locus
 export @genes
 
 include("types.jl")

diff --git a/src/genedataview.jl b/src/genedataview.jl
@@ -1,15 +1,13 @@
 struct GeneDataView{G <: AbstractGene} <: AbstractArray{G, 1}
-    parent::Chromosome{G}
+    parent::Vector{Chromosome{G}}
     indices::Vector{UInt}
     property::Symbol
 end
 
 
 function Base.getproperty(gene::G, name::Symbol) where {G <: AbstractGene}
-    if name in (:parent, :index)
-        return getfield(gene, name)
-    elseif name in propertynames(gene)
-        return gene.parent.genedata[gene.index, name]
+    if name in propertynames(gene)
+        return parent(gene).genedata[index(gene), name]
     else
         return missing
     end
@@ -22,46 +20,42 @@ end
 
 
 function Base.getproperty(genes::AbstractArray{G, 1}, name::Symbol) where {G <: AbstractGene}
-    # @assert all(getfield.(genes, :parent) .== Ref(getfield(genes[1], :parent))) "Not all members of `genes` come from the same parent"
-    if name == :index
-        return [getfield(g, :index) for g in genes]
-    elseif name == :parent
-        return getfield(genes[1], name)
-    else
-        return GeneDataView(genes[1].parent, genes.index, name)
-    end
+    GeneDataView(parent.(genes), getfield.(genes, Ref(:index)), name)
 end
 
 
 function Base.setproperty!(gene::G, name::Symbol, x::T) where {G <: AbstractGene, T}
-    if hasproperty(gene.parent.genedata, name)
-        gene.parent.genedata[gene.index, name] = x
+    if hasproperty(parent(gene).genedata, name)
+        parent(gene).genedata[index(gene), name] = x
     else
-        s = size(gene.parent.genedata, 1)
-        gene.parent.genedata[!, name] = Vector{Union{Missing, T}}(missing, s)
-        gene.parent.genedata[gene.index, name] = x
+        s = size(parent(gene).genedata, 1)
+        parent(gene).genedata[!, name] = Vector{Union{Missing, T}}(missing, s)
+        parent(gene).genedata[index(gene), name] = x
     end
     return x
 end
 
 
 Base.size(gv::GeneDataView) = size(gv.indices)
-Base.getindex(gv::GeneDataView, i::Int) = getproperty(gv.parent.genes[gv.indices[i]], gv.property)
-Base.getindex(gv::GeneDataView, I::AbstractArray) = GeneDataView(gv.parent, gv.indices[I], gv.property)
-Base.setindex!(gv::GeneDataView, v, i::Int) = (Base.setproperty!(gv.parent.genes[i], gv.property, v))
-Base.similar(gv::GeneDataView{Gene}) = GeneDataView(gv.parent, gv.indices, gv.property)
-Base.copy(gv::GeneDataView{Gene}) = GeneDataView(gv.parent, gv.indices, gv.property)
-Base.view(gv::GeneDataView{Gene}, I) = GeneDataView(gv.parent, gv.indices[I], gv.property)
-
-
-function Base.fill!(gv::GeneDataView{Gene}, x)
-    if hasproperty(gv.parent.genedata, gv.property)
-        xT = convert(eltype(gv.parent.genedata[!, gv.property]), x)
-    else
-        xT = x
+Base.getindex(gv::GeneDataView, i::Int) = getproperty(gv.parent[i].genes[gv.indices[i]], gv.property)
+Base.getindex(gv::GeneDataView, I::AbstractArray) = GeneDataView(gv.parent[I], gv.indices[I], gv.property)
+Base.setindex!(gv::GeneDataView, v, i::Int) = (Base.setproperty!(gv.parent[i].genes[gv.indices[i]], gv.property, v))
+Base.similar(gv::GeneDataView{G}) where {G <: AbstractGene} = GeneDataView(gv.parent, gv.indices, gv.property)
+Base.copy(gv::GeneDataView{G}) where {G <: AbstractGene} = GeneDataView(gv.parent, gv.indices, gv.property)
+Base.view(gv::GeneDataView{G}, I) where {G <: AbstractGene} = GeneDataView(gv.parent[I], gv.indices[I], gv.property)
+
+
+function Base.fill!(gv::GeneDataView{G}, x) where {G <: AbstractGene}
+    chrs = unique(gv.parent)
+    for chr in chrs
+        if hasproperty(chr.genedata, gv.property)
+            xT = convert(eltype(chr.genedata[!, gv.property]), x)
+        else
+            xT = x
+        end
     end
-    for I in gv.indices
-        @inbounds gv[I] = x
+    for i in eachindex(gv.indices)
+        @inbounds gv[i] = x
     end
     gv
 end
diff --git a/src/readgbk.jl b/src/readgbk.jl
@@ -9,7 +9,7 @@ end
 
 
 """
-Parse lines encoding genomic position, returning the feature as a `String`, and an instance of `Locus`.
+Parse lines encoding genomic position, returning the feature as a `Symbol`, and an instance of `Locus`.
 """
 function parseposition(line::String)
     feature, posstring = split(strip(line), r" +")
@@ -31,7 +31,7 @@ function parseposition(line::String)
             push!(order, r[1]:r[2])
         end
     end
-    return feature, Locus(position, strand, complete_left, complete_right, order, join)
+    return Symbol(feature), Locus(position, strand, complete_left, complete_right, order, join)
 end
 
 
@@ -49,8 +49,8 @@ end
 
 Parse and return one chromosome entry, and the line number that it ends at.
 """
-function parsechromosome(lines)
-    genes = Gene[]
+function parsechromosome(lines, G::Type = Gene)
+    genes = G[]
     iobuffer = IOBuffer()
 	isheader = true
     isfooter = false
@@ -60,10 +60,10 @@ function parsechromosome(lines)
     content = String("")
     header = ""
 
-    feature = ""
+    feature = :source
     locus = Locus()
 
-    chromosome = Chromosome()
+    chromosome = Chromosome{G}()
 
     linecount = 0
     for line in lines
@@ -120,8 +120,9 @@ function parsechromosome(lines)
                     else
                         (qualifier, content) = match(r"^ +/(\S+)=(\S+)$", line).captures
                         try
-                            content = Meta.parse(content)
-                            content isa Expr && throw(Meta.ParseError)
+                            tmpcontent = Meta.parse(content)
+                            tmpcontent isa Expr && throw(Meta.ParseError)
+							content = tmpcontent
                         catch
                             content = Symbol(content)
                         end
@@ -131,7 +132,9 @@ function parsechromosome(lines)
                         spanning = true
                     end
 
-                    pushproperty!(chromosome.genes[end], Symbol(qualifier), content)
+					isempty(names(chromosome.genedata)) ?
+						(chromosome.genedata[!, Symbol(qualifier)] = Union{Missing, typeof(content)}[content]) :
+	                    pushproperty!(chromosome.genes[end], Symbol(qualifier), content)
 
                 else
                     # Qualifiers without a value assigned to them end up here
@@ -148,7 +151,7 @@ function parsechromosome(lines)
                     spanning = false
                 end
                 if eltype(chromosome.genedata[!, Symbol(qualifier)]).b <: AbstractArray
-                    i = chromosome.genes[end].index
+                    i = index(chromosome.genes[end])
                     chromosome.genedata[!, Symbol(qualifier)][end][end] = Base.getproperty(chromosome.genes[end], Symbol(qualifier))[end] * "\n" * content
                 else
                     Base.setproperty!(chromosome.genes[end], Symbol(qualifier), Base.getproperty(chromosome.genes[end], Symbol(qualifier)) * "\n" * content)
@@ -176,10 +179,10 @@ end
 
 Parse GenBank-formatted file `filename`, returning a `Vector{Chromosome}`.
 """
-function readgbk(filename)
+function readgbk(filename, G::Type = Gene)
     gz = filename[end-2:end] == ".gz"
     finished = false
-    chrs = Chromosome[]
+    chrs = Chromosome{G}[]
     if gz
         f = GZip.open(filename)
     else
@@ -191,7 +194,7 @@ function readgbk(filename)
         if currentline >= length(lines)
             break
         end
-        i, chr = parsechromosome(lines[currentline:end])
+        i, chr = parsechromosome(lines[currentline:end], G)
         currentline += i
         push!(chrs, chr)
     end