mirror of
https://github.com/MillironX/XAM.jl.git
synced 2024-11-26 03:29:55 +00:00
Fun with flags
Implements flag queries.
This commit is contained in:
Ciarán O'Mara
parent
0d1eec3ed3
commit
9a9f2c1f5a
7 changed files with 233 additions and 122 deletions
22
src/XAM.jl
22
src/XAM.jl
|
|
@ -1,6 +1,7 @@
|
||||||
module XAM
|
module XAM
|
||||||
|
|
||||||
using BioGenerics
|
using BioGenerics
|
||||||
|
import BioGenerics: isfilled #Note: used by `ismapped`.
|
||||||
|
|
||||||
export
|
export
|
||||||
SAM,
|
SAM,
|
||||||
|
|
@ -10,26 +11,7 @@ abstract type XAMRecord end
|
||||||
abstract type XAMReader <: BioGenerics.IO.AbstractReader end
|
abstract type XAMReader <: BioGenerics.IO.AbstractReader end
|
||||||
abstract type XAMWriter <: BioGenerics.IO.AbstractWriter end
|
abstract type XAMWriter <: BioGenerics.IO.AbstractWriter end
|
||||||
|
|
||||||
"""
|
include("flags.jl")
|
||||||
flag(record::Union{SAM.Record, BAM.Record})::UInt16
|
|
||||||
|
|
||||||
Get the bitwise flags of `record`. The returned value is a `UInt16` of each flag
|
|
||||||
being OR'd together. The possible flags are:
|
|
||||||
|
|
||||||
0x0001 template having multiple segments in sequencing
|
|
||||||
0x0002 each segment properly aligned according to the aligner
|
|
||||||
0x0004 segment unmapped
|
|
||||||
0x0008 next segment in the template unmapped
|
|
||||||
0x0010 SEQ being reverse complemented
|
|
||||||
0x0020 SEQ of the next segment in the template being reverse complemented
|
|
||||||
0x0040 the first segment in the template
|
|
||||||
0x0080 the last segment in the template
|
|
||||||
0x0100 secondary alignment
|
|
||||||
0x0200 not passing filters, such as platform/vendor quality controls
|
|
||||||
0x0400 PCR or optical duplicate
|
|
||||||
0x0800 supplementary alignment
|
|
||||||
"""
|
|
||||||
function flag end
|
|
||||||
|
|
||||||
include("sam/sam.jl")
|
include("sam/sam.jl")
|
||||||
include("bam/bam.jl")
|
include("bam/bam.jl")
|
||||||
|
|
|
||||||
|
|
@ -6,7 +6,8 @@ module BAM
|
||||||
using BioGenerics
|
using BioGenerics
|
||||||
using GenomicFeatures
|
using GenomicFeatures
|
||||||
using XAM.SAM
|
using XAM.SAM
|
||||||
import ..XAM: flag, XAMRecord, XAMReader, XAMWriter
|
import ..XAM: flag, XAMRecord, XAMReader, XAMWriter,
|
||||||
|
ismapped, isprimary, ispositivestrand, isnextmapped #TODO: Deprecate import of flag queries. These were imported to preseve existing API.
|
||||||
|
|
||||||
import BGZFStreams
|
import BGZFStreams
|
||||||
import BioAlignments
|
import BioAlignments
|
||||||
|
|
|
||||||
|
|
@ -137,37 +137,6 @@ function hasflag(record::Record)
|
||||||
return isfilled(record)
|
return isfilled(record)
|
||||||
end
|
end
|
||||||
|
|
||||||
"""
|
|
||||||
ismapped(record::Record)::Bool
|
|
||||||
|
|
||||||
Test if `record` is mapped.
|
|
||||||
"""
|
|
||||||
function ismapped(record::Record)::Bool
|
|
||||||
return flag(record) & SAM.FLAG_UNMAP == 0
|
|
||||||
end
|
|
||||||
|
|
||||||
"""
|
|
||||||
isprimary(record::Record)::Bool
|
|
||||||
|
|
||||||
Test if `record` is a primary line of the read.
|
|
||||||
|
|
||||||
This is equivalent to `flag(record) & 0x900 == 0`.
|
|
||||||
"""
|
|
||||||
function isprimary(record::Record)::Bool
|
|
||||||
return flag(record) & 0x900 == 0
|
|
||||||
end
|
|
||||||
|
|
||||||
"""
|
|
||||||
ispositivestrand(record::Record)::Bool
|
|
||||||
|
|
||||||
Test if `record` is aligned to the positive strand.
|
|
||||||
|
|
||||||
This is equivalent to `flag(record) & 0x10 == 0`.
|
|
||||||
"""
|
|
||||||
function ispositivestrand(record::Record)::Bool
|
|
||||||
flag(record) & 0x10 == 0
|
|
||||||
end
|
|
||||||
|
|
||||||
"""
|
"""
|
||||||
refid(record::Record)::Int
|
refid(record::Record)::Int
|
||||||
|
|
||||||
|
|
@ -253,15 +222,6 @@ function hasrightposition(record::Record)
|
||||||
return isfilled(record) && ismapped(record)
|
return isfilled(record) && ismapped(record)
|
||||||
end
|
end
|
||||||
|
|
||||||
"""
|
|
||||||
isnextmapped(record::Record)::Bool
|
|
||||||
|
|
||||||
Test if the mate/next read of `record` is mapped.
|
|
||||||
"""
|
|
||||||
function isnextmapped(record::Record)::Bool
|
|
||||||
return isfilled(record) && (flag(record) & SAM.FLAG_MUNMAP == 0)
|
|
||||||
end
|
|
||||||
|
|
||||||
"""
|
"""
|
||||||
nextrefid(record::Record)::Int
|
nextrefid(record::Record)::Int
|
||||||
|
|
||||||
|
|
|
||||||
226
src/flags.jl
Normal file
226
src/flags.jl
Normal file
|
|
@ -0,0 +1,226 @@
|
||||||
|
# Flags
|
||||||
|
# =========
|
||||||
|
#
|
||||||
|
|
||||||
|
"""
|
||||||
|
flag(record::XAMRecord})::UInt16
|
||||||
|
|
||||||
|
Get the bitwise flags of `record`.
|
||||||
|
The returned value is a `UInt16` of each flag being OR'd together.
|
||||||
|
The possible flags are:
|
||||||
|
|
||||||
|
0x0001 template having multiple segments in sequencing
|
||||||
|
0x0002 each segment properly aligned according to the aligner
|
||||||
|
0x0004 segment unmapped
|
||||||
|
0x0008 next segment in the template unmapped
|
||||||
|
0x0010 SEQ being reverse complemented
|
||||||
|
0x0020 SEQ of the next segment in the template being reverse complemented
|
||||||
|
0x0040 the first segment in the template
|
||||||
|
0x0080 the last segment in the template
|
||||||
|
0x0100 secondary alignment
|
||||||
|
0x0200 not passing filters, such as platform/vendor quality controls
|
||||||
|
0x0400 PCR or optical duplicate
|
||||||
|
0x0800 supplementary alignment
|
||||||
|
"""
|
||||||
|
|
||||||
|
function flag end
|
||||||
|
|
||||||
|
# Bitwise flags (or FLAG).
|
||||||
|
for (name, bits, doc) in [
|
||||||
|
(:PAIRED, UInt16(0x001), "the read is paired in sequencing, no matter whether it is mapped in a pair"),
|
||||||
|
(:PROPER_PAIR, UInt16(0x002), "the read is mapped in a proper pair" ),
|
||||||
|
(:UNMAP, UInt16(0x004), "the read itself is unmapped; conflictive with FLAG_PROPER_PAIR" ),
|
||||||
|
(:MUNMAP, UInt16(0x008), "the mate is unmapped" ),
|
||||||
|
(:REVERSE, UInt16(0x010), "the read is mapped to the reverse strand" ),
|
||||||
|
(:MREVERSE, UInt16(0x020), "the mate is mapped to the reverse strand" ),
|
||||||
|
(:READ1, UInt16(0x040), "this is read1" ),
|
||||||
|
(:READ2, UInt16(0x080), "this is read2" ),
|
||||||
|
(:SECONDARY, UInt16(0x100), "not primary alignment" ),
|
||||||
|
(:QCFAIL, UInt16(0x200), "QC failure" ),
|
||||||
|
(:DUP, UInt16(0x400), "optical or PCR duplicate" ),
|
||||||
|
(:SUPPLEMENTARY, UInt16(0x800), "supplementary alignment" ),]
|
||||||
|
@assert bits isa UInt16 "The bits must be of type UInt16."
|
||||||
|
sym = Symbol("FLAG_", name)
|
||||||
|
docstring = """ $sym
|
||||||
|
SAM/BAM flag: $doc
|
||||||
|
|
||||||
|
See also: [`flag`](@ref)
|
||||||
|
"""
|
||||||
|
@eval begin
|
||||||
|
@doc $(docstring) const $(sym) = $(bits)
|
||||||
|
end
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
ispaired(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record`'s template has multiple segments in sequencing.
|
||||||
|
"""
|
||||||
|
function ispaired(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_PAIRED == FLAG_PAIRED
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
isproperpair(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether each segment of the `record`'s template properly aligned according to the aligner.
|
||||||
|
"""
|
||||||
|
function isproperpair(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & PROPER_PAIR == PROPER_PAIR
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
isunmapped(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record` is unmapped.
|
||||||
|
"""
|
||||||
|
function isunmapped(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_UNMAP == FLAG_UNMAP
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
ismapped(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record` is mapped.
|
||||||
|
"""
|
||||||
|
function ismapped(record::XAMRecord)::Bool
|
||||||
|
# return flag(record) & FLAG_UNMAP == 0
|
||||||
|
return isfilled(record) && (flag(record) & FLAG_UNMAP == 0)
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
ismateunmapped(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record`'s mate is unmapped.
|
||||||
|
"""
|
||||||
|
function ismateunmapped(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_MUNMAP == FLAG_MUNMAP
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
isnextmapped(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Test if the mate/next read of `record` is mapped.
|
||||||
|
"""
|
||||||
|
function isnextmapped(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_MUNMAP == 0
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
isreverse(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record` is mapped to the reverse strand.
|
||||||
|
"""
|
||||||
|
function isreverse(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_REVERSE == FLAG_REVERSE
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
isforward(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record` is mapped to the forward strand.
|
||||||
|
"""
|
||||||
|
function isforward(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_REVERSE == 0
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
ispositivestrand(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether `record` is aligned to the positive strand.
|
||||||
|
"""
|
||||||
|
function ispositivestrand(record::XAMRecord)::Bool
|
||||||
|
return isforward(record)
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
ispositivestrand(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether `record` is aligned to the negative strand.
|
||||||
|
"""
|
||||||
|
function isnegativestrand(record::XAMRecord)::Bool
|
||||||
|
return isreverse(record)
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
ismatereverse(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record`'s mate is mapped to the reverse strand.
|
||||||
|
"""
|
||||||
|
function ismatereverse(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_MREVERSE == FLAG_MREVERSE
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
isread1(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record` is read1.
|
||||||
|
"""
|
||||||
|
function isread1(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_READ1 == FLAG_READ1
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
isread2(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record` is read2.
|
||||||
|
"""
|
||||||
|
function isread2(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_READ2 == FLAG_READ2
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
issecondaryalignment(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record` is a secondary alignment.
|
||||||
|
"""
|
||||||
|
function issecondaryalignment(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_SECONDARY == FLAG_SECONDARY
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
isprimaryalignment(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record` is the primary alignment.
|
||||||
|
"""
|
||||||
|
function isprimaryalignment(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_SECONDARY == 0
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
isqcfail(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record` did not pass filters, such as platform/vendor quality controls.
|
||||||
|
"""
|
||||||
|
function isqcfail(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_QCFAIL == FLAG_QCFAIL
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
isduplicate(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record` is a PCR or optical duplicate.
|
||||||
|
"""
|
||||||
|
function isduplicate(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_DUP == FLAG_DUP
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
issupplementaryalignment(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether the `record` is a supplementary alignment.
|
||||||
|
"""
|
||||||
|
function issupplementaryalignment(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & FLAG_SUPPLEMENTARY == FLAG_SUPPLEMENTARY
|
||||||
|
end
|
||||||
|
|
||||||
|
"""
|
||||||
|
isprimary(record::XAMRecord)::Bool
|
||||||
|
|
||||||
|
Query whether `record` is a primary line of the read.
|
||||||
|
|
||||||
|
This is equivalent to `flag(record) & 0x900 == 0`.
|
||||||
|
"""
|
||||||
|
function isprimary(record::XAMRecord)::Bool
|
||||||
|
return flag(record) & 0x900 == 0
|
||||||
|
end
|
||||||
|
|
@ -1,29 +0,0 @@
|
||||||
# SAM Flags
|
|
||||||
# =========
|
|
||||||
#
|
|
||||||
# Bitwise flags (or FLAG).
|
|
||||||
|
|
||||||
for (name, bits, doc) in [
|
|
||||||
(:PAIRED, UInt16(0x001), "the read is paired in sequencing, no matter whether it is mapped in a pair"),
|
|
||||||
(:PROPER_PAIR, UInt16(0x002), "the read is mapped in a proper pair" ),
|
|
||||||
(:UNMAP, UInt16(0x004), "the read itself is unmapped; conflictive with SAM.FLAG_PROPER_PAIR" ),
|
|
||||||
(:MUNMAP, UInt16(0x008), "the mate is unmapped" ),
|
|
||||||
(:REVERSE, UInt16(0x010), "the read is mapped to the reverse strand" ),
|
|
||||||
(:MREVERSE, UInt16(0x020), "the mate is mapped to the reverse strand" ),
|
|
||||||
(:READ1, UInt16(0x040), "this is read1" ),
|
|
||||||
(:READ2, UInt16(0x080), "this is read2" ),
|
|
||||||
(:SECONDARY, UInt16(0x100), "not primary alignment" ),
|
|
||||||
(:QCFAIL, UInt16(0x200), "QC failure" ),
|
|
||||||
(:DUP, UInt16(0x400), "optical or PCR duplicate" ),
|
|
||||||
(:SUPPLEMENTARY, UInt16(0x800), "supplementary alignment" ),]
|
|
||||||
@assert bits isa UInt16 "The bits must be of type UInt16."
|
|
||||||
sym = Symbol("FLAG_", name)
|
|
||||||
docstring = """ $sym
|
|
||||||
SAM/BAM flag: $doc
|
|
||||||
|
|
||||||
See also: [`flag`](@ref)
|
|
||||||
"""
|
|
||||||
@eval begin
|
|
||||||
@doc $(docstring) const $(sym) = $(bits)
|
|
||||||
end
|
|
||||||
end
|
|
||||||
|
|
@ -159,26 +159,6 @@ function hasflag(record::Record)
|
||||||
return isfilled(record)
|
return isfilled(record)
|
||||||
end
|
end
|
||||||
|
|
||||||
"""
|
|
||||||
ismapped(record::Record)::Bool
|
|
||||||
|
|
||||||
Test if `record` is mapped.
|
|
||||||
"""
|
|
||||||
function ismapped(record::Record)::Bool
|
|
||||||
return isfilled(record) && (flag(record) & FLAG_UNMAP == 0)
|
|
||||||
end
|
|
||||||
|
|
||||||
"""
|
|
||||||
isprimary(record::Record)::Bool
|
|
||||||
|
|
||||||
Test if `record` is a primary line of the read.
|
|
||||||
|
|
||||||
This is equivalent to `flag(record) & 0x900 == 0`.
|
|
||||||
"""
|
|
||||||
function isprimary(record::Record)::Bool
|
|
||||||
return flag(record) & 0x900 == 0
|
|
||||||
end
|
|
||||||
|
|
||||||
"""
|
"""
|
||||||
refname(record::Record)::String
|
refname(record::Record)::String
|
||||||
|
|
||||||
|
|
@ -227,15 +207,6 @@ function hasrightposition(record::Record)
|
||||||
return hasposition(record) && hasalignment(record)
|
return hasposition(record) && hasalignment(record)
|
||||||
end
|
end
|
||||||
|
|
||||||
"""
|
|
||||||
isnextmapped(record::Record)::Bool
|
|
||||||
|
|
||||||
Test if the mate/next read of `record` is mapped.
|
|
||||||
"""
|
|
||||||
function isnextmapped(record::Record)::Bool
|
|
||||||
return isfilled(record) && (flag(record) & FLAG_MUNMAP == 0)
|
|
||||||
end
|
|
||||||
|
|
||||||
"""
|
"""
|
||||||
nextrefname(record::Record)::String
|
nextrefname(record::Record)::String
|
||||||
|
|
||||||
|
|
|
||||||
|
|
@ -11,7 +11,8 @@ import BioGenerics.Exceptions: missingerror
|
||||||
import BioGenerics.Automa: State
|
import BioGenerics.Automa: State
|
||||||
import BioSequences
|
import BioSequences
|
||||||
import TranscodingStreams: TranscodingStreams, TranscodingStream
|
import TranscodingStreams: TranscodingStreams, TranscodingStream
|
||||||
import ..XAM: flag, XAMRecord, XAMReader, XAMWriter
|
import ..XAM: flag, XAMRecord, XAMReader, XAMWriter,
|
||||||
|
ismapped, isprimary, ispositivestrand, isnextmapped #TODO: Deprecate import of flag queries. These were imported to preseve existing API.
|
||||||
|
|
||||||
using Printf: @sprintf
|
using Printf: @sprintf
|
||||||
|
|
||||||
|
|
@ -48,7 +49,6 @@ function unsafe_parse_decimal(::Type{T}, data::Vector{UInt8}, range::UnitRange{I
|
||||||
end
|
end
|
||||||
|
|
||||||
|
|
||||||
include("flags.jl")
|
|
||||||
include("metainfo.jl")
|
include("metainfo.jl")
|
||||||
include("record.jl")
|
include("record.jl")
|
||||||
include("header.jl")
|
include("header.jl")
|
||||||
|
|
|
||||||
Loading…
Reference in a new issue