DNAシーケンシング技術の急速な低下に伴い、デノボ全ゲノムシーケンシング（WGS）プロジェクトは新しいゲノムについて非常に深いカバレッジを生み出している。しかし、これらの技術による高いカバレッジとゲノムアセンブリアルゴリズム（Gnerre et al、2011; Li et al、2010）の劇的な改善を持ってしても、次世代技術によって生み出されるリード長の短さは、 WGSデータからゲノムを再構築する上で重要な障壁となっている。リード長が増加すると、ゲノムアセンブリの品質に重大かつ積極的な影響がある。
現在のいくつかのプロトコルは、DNA断片ライブラリーの両端から配列を生成する。フラグメントがシーケンスできる長さの2倍よりも短い場合、結果として得られるペアエンドリードにはオーバーラップが存在することになる。たとえば、プロジェクトでは100 bpのリード、175 bpのライブラリを使用する場合がある。このタイプのライブラリは、アセンブリでこれらのリードを使用する前に、ペアエンドのリードの末端同士をオーバーラップおよびマージすることによってリード長を延長する機会が提供される。
この論文では、FLASH（Fast Length Adjustment of SHort reads）を紹介する。これは、ペアエンドリード間の正確なオーバーラップを検出し、それらをつなぎ合わせてリードを拡張するソフトウェアツールである。著者らはシミュレーションで100万の100 bpペアエンドリードを発生させ、ツールの正確性をテストし、信頼性が高いと判断した。次に、実際のデータを使用して、このマージ手順がアセンブリの品質に及ぼす影響を測定した。また、FLASHの速度と精度をSHERA（Rodrigue et al、2010）と比較した。
リードのエラー率が1％以下の場合、FLASHはペアリードの99％を正しく処理できる。エラー率が2％の場合、FLASHはペアリードの98％以上を処理する（デフォルトパラメータ）。より積極的なパラメータ（すなわち、-x 0.35）では、エラー率は5％でもFLASHはペアリードの90％を正しく処理する。
Limitations of FLASH include:
- FLASH cannot merge paired-end reads that do not overlap.
- FLASH is not designed for data that has a significant amount of indel
errors (such as Sanger sequencing data). It is best suited for Illumina
data.
flash2 Githubより
インストール
mac os10.14の miniconda3-4.0.5環境でテストした。
本体 Github
#Anaconda環境でcondaを使い導入
conda install -c bioconda -y flash2
> flash2 -h
$ flash2 -h
Usage: flash [OPTIONS] MATES_1.FASTQ MATES_2.FASTQ
flash [OPTIONS] --interleaved-input (MATES.FASTQ | -)
flash [OPTIONS] --tab-delimited-input (MATES.TAB | -)
----------------------------------------------------------------------------
DESCRIPTION
----------------------------------------------------------------------------
FLASH (Fast Length Adjustment of SHort reads) is an accurate and fast tool
to merge paired-end reads that were generated from DNA fragments whose
lengths are shorter than twice the length of reads. Merged read pairs result
in unpaired longer reads, which are generally more desired in genome
assembly and genome analysis processes.
Briefly, the FLASH algorithm considers all possible overlaps at or above a
minimum length between the reads in a pair and chooses the overlap that
results in the lowest mismatch density (proportion of mismatched bases in
the overlapped region). Ties between multiple overlaps are broken by
considering quality scores at mismatch sites. When building the merged
sequence, FLASH computes a consensus sequence in the overlapped region.
More details can be found in the original publication
(http://bioinformatics.oxfordjournals.org/content/27/21/2957.full).
Limitations of FLASH include:
- FLASH cannot merge paired-end reads that do not overlap.
- FLASH is not designed for data that has a significant amount of indel
errors (such as Sanger sequencing data). It is best suited for Illumina
data.
----------------------------------------------------------------------------
MANDATORY INPUT
----------------------------------------------------------------------------
The most common input to FLASH is two FASTQ files containing read 1 and read 2
of each mate pair, respectively, in the same order.
Alternatively, you may provide one FASTQ file, which may be standard input,
containing paired-end reads in either interleaved FASTQ (see the
--interleaved-input option) or tab-delimited (see the --tab-delimited-input
option) format. In all cases, gzip compressed input is autodetected. Also,
in all cases, the PHRED offset is, by default, assumed to be 33; use the
--phred-offset option to change it.
----------------------------------------------------------------------------
OUTPUT
----------------------------------------------------------------------------
The default output of FLASH consists of the following files:
- out.extendedFrags.fastq The merged reads.
- out.notCombined_1.fastq Read 1 of mate pairs that were not merged.
- out.notCombined_2.fastq Read 2 of mate pairs that were not merged.
- out.hist Numeric histogram of merged read lengths.
- out.histogram Visual histogram of merged read lengths.
FLASH also logs informational messages to standard output. These can also be
redirected to a file, as in the following example:
$ flash reads_1.fq reads_2.fq 2>&1 | tee flash.log
In addition, FLASH supports several features affecting the output:
- Writing the merged reads directly to standard output (--to-stdout)
- Writing gzip compressed output files (-z) or using an external
compression program (--compress-prog)
- Writing the uncombined read pairs in interleaved FASTQ format
(--interleaved-output)
- Writing all output reads to a single file in tab-delimited format
(--tab-delimited-output)
----------------------------------------------------------------------------
OPTIONS
----------------------------------------------------------------------------
-Q, --quality-cutoff=NUM The cut off number for the quality score
corresponding wtih the percent cutoff. Default:
2.
-C, --percent-cutoff=NUM The cutoff percentage for each read that will
be discarded if it falls below -Q option. (0-100) Default:
50.
-D, --no-discard This turns off the discard logic Default: false
-m, --min-overlap=NUM The minimum required overlap length between two
reads to provide a confident overlap. Default 10bp.
-M, --max-overlap=NUM Maximum overlap length expected in approximately
90% of read pairs. It is by default set to 65bp,
which works well for 100bp reads generated from a
180bp library, assuming a normal distribution of
fragment lengths. Overlaps longer than the maximum
overlap parameter are still considered as good
overlaps, but the mismatch density (explained below)
is calculated over the first max_overlap bases in
the overlapped region rather than the entire
overlap. Default: 65bp, or calculated from the
specified read length, fragment length, and fragment
length standard deviation.
-e, --min-overlap-outie=NUM The minimum required overlap length between two
reads to provide a confident overlap in an outie scenario.
Default: 35bp.
-x, --max-mismatch-density=NUM
Maximum allowed ratio between the number of
mismatched base pairs and the overlap length.
Two reads will not be combined with a given overlap
if that overlap results in a mismatched base density
higher than this value. Note: Any occurence of an
'N' in either read is ignored and not counted
towards the mismatches or overlap length. Our
experimental results suggest that higher values of
the maximum mismatch density yield larger
numbers of correctly merged read pairs but at
the expense of higher numbers of incorrectly
merged read pairs. Default: 0.25.
-O, --allow-outies Also try combining read pairs in the "outie"
orientation, e.g.
Read 1: <-----------
Read 2: ------------>
as opposed to only the "innie" orientation, e.g.
Read 1: <------------
Read 2: ----------->
FLASH uses the same parameters when trying each
orientation. If a read pair can be combined in
both "innie" and "outie" orientations, the
better-fitting one will be chosen using the same
scoring algorithm that FLASH normally uses.
This option also causes extra .innie and .outie
histogram files to be produced.
-p, --phred-offset=OFFSET
The smallest ASCII value of the characters used to
represent quality values of bases in FASTQ files.
It should be set to either 33, which corresponds
to the later Illumina platforms and Sanger
platforms, or 64, which corresponds to the
earlier Illumina platforms. Default: 33.
-r, --read-len=LEN
-f, --fragment-len=LEN
-s, --fragment-len-stddev=LEN
Average read length, fragment length, and fragment
standard deviation. These are convenience parameters
only, as they are only used for calculating the
maximum overlap (--max-overlap) parameter.
The maximum overlap is calculated as the overlap of
average-length reads from an average-size fragment
plus 2.5 times the fragment length standard
deviation. The default values are -r 100, -f 180,
and -s 18, so this works out to a maximum overlap of
65 bp. If --max-overlap is specified, then the
specified value overrides the calculated value.
If you do not know the standard deviation of the
fragment library, you can probably assume that the
standard deviation is 10% of the average fragment
length.
--cap-mismatch-quals Cap quality scores assigned at mismatch locations
to 2. This was the default behavior in FLASH v1.2.7
and earlier. Later versions will instead calculate
such scores as max(|q1 - q2|, 2); that is, the
absolute value of the difference in quality scores,
but at least 2. Essentially, the new behavior
prevents a low quality base call that is likely a
sequencing error from significantly bringing down
the quality of a high quality, likely correct base
call.
--interleaved-input Instead of requiring files MATES_1.FASTQ and
MATES_2.FASTQ, allow a single file MATES.FASTQ that
has the paired-end reads interleaved. Specify "-"
to read from standard input.
--interleaved-output Write the uncombined pairs in interleaved FASTQ
format.
-I, --interleaved Equivalent to specifying both --interleaved-input
and --interleaved-output.
-Ti, --tab-delimited-input
Assume the input is in tab-delimited format
rather than FASTQ, in the format described below in
'--tab-delimited-output'. In this mode you should
provide a single input file, each line of which must
contain either a read pair (5 fields) or a single
read (3 fields). FLASH will try to combine the read
pairs. Single reads will be written to the output
file as-is if also using --tab-delimited-output;
otherwise they will be ignored. Note that you may
specify "-" as the input file to read the
tab-delimited data from standard input.
-To, --tab-delimited-output
Write output in tab-delimited format (not FASTQ).
Each line will contain either a combined pair in the
format 'tag <tab> seq <tab> qual' or an uncombined
pair in the format 'tag <tab> seq_1 <tab> qual_1
<tab> seq_2 <tab> qual_2'.
-o, --output-prefix=PREFIX
Prefix of output files. Default: "out".
-d, --output-directory=DIR
Path to directory for output files. Default:
current working directory.
-c, --to-stdout Write the combined reads to standard output. In
this mode, with FASTQ output (the default) the
uncombined reads are discarded. With tab-delimited
output, uncombined reads are included in the
tab-delimited data written to standard output.
In both cases, histogram files are not written,
and informational messages are sent to standard
error rather than to standard output.
-z, --compress Compress the output files directly with zlib,
using the gzip container format. Similar to
specifying --compress-prog=gzip and --suffix=gz,
but may be slightly faster.
--compress-prog=PROG Pipe the output through the compression program
PROG, which will be called as `PROG -c -',
plus any arguments specified by --compress-prog-args.
PROG must read uncompressed data from standard input
and write compressed data to standard output when
invoked as noted above.
Examples: gzip, bzip2, xz, pigz.
--compress-prog-args=ARGS
A string of additional arguments that will be passed
to the compression program if one is specified with
--compress-prog=PROG. (The arguments '-c -' are
still passed in addition to explicitly specified
arguments.)
--suffix=SUFFIX, --output-suffix=SUFFIX
Use SUFFIX as the suffix of the output files
after ".fastq". A dot before the suffix is assumed,
unless an empty suffix is provided. Default:
nothing; or 'gz' if -z is specified; or PROG if
--compress-prog=PROG is specified.
-t, --threads=NTHREADS Set the number of worker threads. This is in
addition to the I/O threads. Default: number of
processors. Note: if you need FLASH's output to
appear deterministically or in the same order as
the original reads, you must specify -t 1
(--threads=1).
-q, --quiet Do not print informational messages.
-h, --help Display this help and exit.
-v, --version Display version.
Run `flash2 --help | less' to prevent this text from scrolling by.
実行方法
ペアエンドのfastq
flash2 pair1.fq pair2.fq -t 8
interleaveのfastq
flash2 --interleaved-input interleave.fq -t 8
リード長のヒストグラムファイル、マージされたリードのfastq、マージされなかったリードのfastqがそれぞれ出力される。
-O (--allow-outies)をつけることで、外向きのペアエンドfastqのマージも可能
flash2 --allow-outies pair1.fq pair2.fq -t 8
-O, --allow-outies Also try combining read pairs in the "outie"
orientation, e.g.
Read 1: <-----------
Read 2: ------------>
as opposed to only the "innie" orientation, e.g.
Read 1: <------------
Read 2: ----------->
FLASH uses the same parameters when trying each
orientation. If a read pair can be combined in
both "innie" and "outie" orientations, the
better-fitting one will be chosen using the same
scoring algorithm that FLASH normally uses.
This option also causes extra .innie and .outie
histogram files to be produced.
引用
FLASH: fast length adjustment of short reads to improve genome assemblies
Tanja Magoč, Steven L. Salzberg
Bioinformatics, Volume 27, Issue 21, 1 November 2011, Pages 2957–2963,
