|
|
|
@ -1,41 +1,43 @@
|
|
|
|
"""
|
|
|
|
"""
|
|
|
|
Variant{S<:BioSequence,T<:BioSymbol}
|
|
|
|
Haplotype{S<:BioSequence,T<:BioSymbol}
|
|
|
|
|
|
|
|
|
|
|
|
A set of variations within a given sequence that are all found together. Depending on the
|
|
|
|
A set of variations within a given sequence that are all found together. Depending on the
|
|
|
|
field, it might also be referred to as a "genotype," "haplotype," or "strain."
|
|
|
|
field, it might also be referred to as a "genotype," "haplotype," or "strain."
|
|
|
|
|
|
|
|
|
|
|
|
# Constructors
|
|
|
|
# Constructors
|
|
|
|
|
|
|
|
|
|
|
|
Variant(ref::S, edits::Vector{Edit{S,T}}) where {S<:BioSequence,T<:BioSymbol}
|
|
|
|
Haplotype(ref::S, edits::Vector{Edit{S,T}}) where {S<:BioSequence,T<:BioSymbol}
|
|
|
|
Variant(ref::S, vars::Vector{Variation{S,T}}) where {S<:BioSequence,T<:BioSymbol}
|
|
|
|
Haplotype(ref::S, vars::Vector{Variation{S,T}}) where {S<:BioSequence,T<:BioSymbol}
|
|
|
|
Variant(
|
|
|
|
Haplotype(
|
|
|
|
aln::PairwiseAlignment{T,T}
|
|
|
|
aln::PairwiseAlignment{T,T}
|
|
|
|
) where {T<:LongSequence{<:Union{BS.AminoAcidAlphabet,BS.NucleicAcidAlphabet}}}
|
|
|
|
) where {T<:LongSequence{<:Union{BS.AminoAcidAlphabet,BS.NucleicAcidAlphabet}}}
|
|
|
|
|
|
|
|
|
|
|
|
When constructing a `Variant` from a vector of [`Edit`](@ref)s or [`Variation`](@ref)s, the
|
|
|
|
When constructing a `Haplotype` from a vector of [`Edit`](@ref)s or [`Variation`](@ref)s,
|
|
|
|
edits are applied sequentially from first to last position, therefore the vector must always
|
|
|
|
the edits are applied sequentially from first to last position, therefore the vector must
|
|
|
|
be sorted by position. These edits are sorted automatically if constructing from an
|
|
|
|
always be sorted by position. These edits are sorted automatically if constructing from an
|
|
|
|
alignment.
|
|
|
|
alignment.
|
|
|
|
"""
|
|
|
|
"""
|
|
|
|
struct Variant{S<:BioSequence,T<:BioSymbol}
|
|
|
|
struct Haplotype{S<:BioSequence,T<:BioSymbol}
|
|
|
|
ref::S
|
|
|
|
ref::S
|
|
|
|
edits::Vector{Edit{S,T}}
|
|
|
|
edits::Vector{Edit{S,T}}
|
|
|
|
|
|
|
|
|
|
|
|
Variant{S,T}(ref::S, edits::Vector{Edit{S,T}}, ::Unsafe) where {S,T} = new(ref, edits)
|
|
|
|
Haplotype{S,T}(ref::S, edits::Vector{Edit{S,T}}, ::Unsafe) where {S,T} = new(ref, edits)
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
|
|
|
|
function Variant{S,T}(ref::S, edits::Vector{Edit{S,T}}) where {S<:BioSequence,T<:BioSymbol}
|
|
|
|
function Haplotype{S,T}(
|
|
|
|
|
|
|
|
ref::S, edits::Vector{Edit{S,T}}
|
|
|
|
|
|
|
|
) where {S<:BioSequence,T<:BioSymbol}
|
|
|
|
sort!(edits; by=x -> x.pos)
|
|
|
|
sort!(edits; by=x -> x.pos)
|
|
|
|
result = Variant{S,T}(ref, edits, Unsafe())
|
|
|
|
result = Haplotype{S,T}(ref, edits, Unsafe())
|
|
|
|
_is_valid(result) || error("TODO") # report what kind of error message?
|
|
|
|
_is_valid(result) || error("TODO") # report what kind of error message?
|
|
|
|
return result
|
|
|
|
return result
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
|
|
|
|
function Variant(ref::S, edits::Vector{Edit{S,T}}) where {S<:BioSequence,T<:BioSymbol}
|
|
|
|
function Haplotype(ref::S, edits::Vector{Edit{S,T}}) where {S<:BioSequence,T<:BioSymbol}
|
|
|
|
return Variant{S,T}(ref, edits)
|
|
|
|
return Haplotype{S,T}(ref, edits)
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
|
|
|
|
function Base.show(io::IO, x::Variant)
|
|
|
|
function Base.show(io::IO, x::Haplotype)
|
|
|
|
n = length(x.edits)
|
|
|
|
n = length(x.edits)
|
|
|
|
print(io, summary(x), " with $n edit$(n > 1 ? "s" : ""):")
|
|
|
|
print(io, summary(x), " with $n edit$(n > 1 ? "s" : ""):")
|
|
|
|
for i in x.edits
|
|
|
|
for i in x.edits
|
|
|
|
@ -46,12 +48,12 @@ function Base.show(io::IO, x::Variant)
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
|
|
|
|
"""
|
|
|
|
"""
|
|
|
|
is_valid(v::Variant)
|
|
|
|
is_valid(v::Haplotype)
|
|
|
|
|
|
|
|
|
|
|
|
Validate `v`. `v` is invalid if any of its operations are out of bounds, or the same
|
|
|
|
Validate `v`. `v` is invalid if any of its operations are out of bounds, or the same
|
|
|
|
position is affected by multiple edits.
|
|
|
|
position is affected by multiple edits.
|
|
|
|
"""
|
|
|
|
"""
|
|
|
|
function _is_valid(v::Variant)
|
|
|
|
function _is_valid(v::Haplotype)
|
|
|
|
isempty(v.ref) && return false
|
|
|
|
isempty(v.ref) && return false
|
|
|
|
valid_positions = 1:length(v.ref)
|
|
|
|
valid_positions = 1:length(v.ref)
|
|
|
|
last_was_insert = false
|
|
|
|
last_was_insert = false
|
|
|
|
@ -87,7 +89,7 @@ function _is_valid(v::Variant)
|
|
|
|
return true
|
|
|
|
return true
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
|
|
|
|
function Variant(
|
|
|
|
function Haplotype(
|
|
|
|
aln::PairwiseAlignment{T,T}
|
|
|
|
aln::PairwiseAlignment{T,T}
|
|
|
|
) where {T<:LongSequence{<:Union{BS.AminoAcidAlphabet,BS.NucleicAcidAlphabet}}}
|
|
|
|
) where {T<:LongSequence{<:Union{BS.AminoAcidAlphabet,BS.NucleicAcidAlphabet}}}
|
|
|
|
ref = aln.b
|
|
|
|
ref = aln.b
|
|
|
|
@ -141,30 +143,30 @@ function Variant(
|
|
|
|
end
|
|
|
|
end
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
|
|
|
|
return Variant(ref, edits)
|
|
|
|
return Haplotype(ref, edits)
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
|
|
|
|
"""
|
|
|
|
"""
|
|
|
|
_edits(v::Variant)
|
|
|
|
_edits(v::Haplotype)
|
|
|
|
|
|
|
|
|
|
|
|
Gets the [`Edit`](@ref)s that comprise `v`
|
|
|
|
Gets the [`Edit`](@ref)s that comprise `v`
|
|
|
|
"""
|
|
|
|
"""
|
|
|
|
_edits(v::Variant) = v.edits
|
|
|
|
_edits(v::Haplotype) = v.edits
|
|
|
|
|
|
|
|
|
|
|
|
"""
|
|
|
|
"""
|
|
|
|
reference(v::Variant)
|
|
|
|
reference(v::Haplotype)
|
|
|
|
|
|
|
|
|
|
|
|
Gets the reference sequence of `v`.
|
|
|
|
Gets the reference sequence of `v`.
|
|
|
|
"""
|
|
|
|
"""
|
|
|
|
reference(v::Variant) = v.ref
|
|
|
|
reference(v::Haplotype) = v.ref
|
|
|
|
Base.:(==)(x::Variant, y::Variant) = x.ref == y.ref && x.edits == y.edits
|
|
|
|
Base.:(==)(x::Haplotype, y::Haplotype) = x.ref == y.ref && x.edits == y.edits
|
|
|
|
|
|
|
|
|
|
|
|
"""
|
|
|
|
"""
|
|
|
|
reconstruct!(seq::S, x::Variant{S}) where {S}
|
|
|
|
reconstruct!(seq::S, x::Haplotype{S}) where {S}
|
|
|
|
|
|
|
|
|
|
|
|
Apply the edits in `x` to `seq` and return the mutated sequence
|
|
|
|
Apply the edits in `x` to `seq` and return the mutated sequence
|
|
|
|
"""
|
|
|
|
"""
|
|
|
|
function reconstruct!(seq::S, x::Variant{S}) where {S}
|
|
|
|
function reconstruct!(seq::S, x::Haplotype{S}) where {S}
|
|
|
|
len = length(x.ref) + sum(edit -> _lendiff(edit), _edits(x))
|
|
|
|
len = length(x.ref) + sum(edit -> _lendiff(edit), _edits(x))
|
|
|
|
resize!(seq, len % UInt)
|
|
|
|
resize!(seq, len % UInt)
|
|
|
|
refpos = seqpos = 1
|
|
|
|
refpos = seqpos = 1
|
|
|
|
|
