~ubuntu-branches/ubuntu/vivid/idba/vivid

« back to all changes in this revision

Viewing changes to src/tools/sort_reads.cpp

Committer: Package Import Robot
Author(s): Andreas Tille
Date: 2014-02-13 13:57:55 UTC
Revision ID: package-import@ubuntu.com-20140213135755-c97jpl9wl5yzfwfw

Tags: upstream-1.1.1

Import upstream version 1.1.1

files added:

AUTHORS

ChangeLog

Makefile.am

Makefile.in

NEWS

README

aclocal.m4

bin/Makefile.am

bin/Makefile.in

config.h.in

configure

configure.ac

depcomp

install-sh

lib/Makefile.am

lib/Makefile.in

missing

script

script/Makefile.am

script/Makefile.in

script/run-unittest.py

script/scan.py

script/validate_blat

script/validate_blat_parallel

src/assembly

src/assembly/assembly_utility.cpp

src/assembly/assembly_utility.h

src/assembly/local_assembler.cpp

src/assembly/local_assembler.h

src/basic

src/basic/atomic_integer.h

src/basic/bit_operation.h

src/basic/functional.h

src/basic/hash.h

src/basic/histgram.h

src/basic/kmer.h

src/basic/math.cpp

src/basic/math.h

src/basic/pool.h

src/container

src/container/hash_map.h

src/container/hash_set.h

src/container/hash_table.h

src/container/managed_list.h

src/graph

src/graph/bit_edges.h

src/graph/contig_builder.h

src/graph/contig_graph.cpp

src/graph/contig_graph.h

src/graph/contig_graph_branch_group.cpp

src/graph/contig_graph_branch_group.h

src/graph/contig_graph_path.h

src/graph/contig_graph_vertex.h

src/graph/contig_info.cpp

src/graph/contig_info.h

src/graph/hash_graph.cpp

src/graph/hash_graph.h

src/graph/hash_graph_branch_group.cpp

src/graph/hash_graph_branch_group.h

src/graph/hash_graph_path.h

src/graph/hash_graph_vertex.h

src/graph/scaffold_graph.cpp

src/graph/scaffold_graph.h

src/graph/scaffold_graph_path.h

src/graph/scaffold_graph_vertex.h

src/graph/vertex_status.h

src/misc

src/misc/blat_record.cpp

src/misc/blat_record.h

src/misc/hash_aligner.cpp

src/misc/hash_aligner.h

src/misc/log.cpp

src/misc/log.h

src/misc/options_description.cpp

src/misc/options_description.h

src/misc/utils.cpp

src/misc/utils.h

src/release

src/release/filterfa.cpp

src/release/fq2fa.cpp

src/release/idba.cpp

src/release/idba_hybrid.cpp

src/release/idba_tran.cpp

src/release/idba_ud.cpp

src/release/parallel_blat.cpp

src/release/raw_n50.cpp

src/release/sim_reads.cpp

src/release/sim_reads_tran.cpp

src/release/split_scaffold.cpp

src/release/validate_contigs_blat.cpp

src/sequence

src/sequence/compact_sequence.cpp

src/sequence/compact_sequence.h

src/sequence/read_library.h

src/sequence/sequence.cpp

src/sequence/sequence.h

src/sequence/sequence_io.cpp

src/sequence/sequence_io.h

src/sequence/sequence_reader.cpp

src/sequence/sequence_reader.h

src/sequence/sequence_writer.cpp

src/sequence/sequence_writer.h

src/sequence/short_sequence.cpp

src/sequence/short_sequence.h

src/tools

src/tools/fa2fq.cpp

src/tools/filter_blat.cpp

src/tools/filter_contigs.cpp

src/tools/idba_tran_test.cpp

src/tools/parallel_rna_blat.cpp

src/tools/print_graph.cpp

src/tools/sample_reads.cpp

src/tools/scaffold.cpp

src/tools/shuffle_reads.cpp

src/tools/sort_psl.cpp

src/tools/sort_reads.cpp

src/tools/split_fa.cpp

src/tools/split_fq.cpp

src/tools/test.cpp

src/tools/validate_component.cpp

src/tools/validate_contigs_mummer.cpp

src/tools/validate_reads_blat.cpp

src/tools/validate_rna.cpp

test

test/Makefile.am

test/Makefile.in

Show diffs side-by-side

added added

removed removed

src/tools/sort_reads.cpp

/**

* @file fq2fa.cpp

* @brief

* @author Yu Peng (ypeng@cs.hku.hk)

* @version 1.0.6

* @date 2011-10-31

#include <algorithm>

#include <cctype>

#include <cstdio>

#include <cstring>

#include <deque>

#include <iostream>

#include <stdexcept>

#include "misc/options_description.h"

#include "misc/utils.h"

#include "sequence/sequence.h"

#include "sequence/sequence_io.h"

#include "sequence/short_sequence.h"

using namespace std;

bool is_paired = false;

bool is_merged = false;

bool is_filtered = false;

int min_length = 0;

deque<ShortSequence> reads;

vector<int> aux;

bool Compare(int x, int y)

{

if (reads[x] != reads[y])

return reads[x] < reads[y];

else

return reads[x+1] < reads[y+1];

}

int main(int argc, char *argv[])

{

OptionsDescription desc;

desc.AddOption("paired", "", is_paired, "if the reads are paired-end in one file");

desc.AddOption("merge", "", is_merged, "if the reads are paired-end in two files");

desc.AddOption("filter", "", is_filtered, "filter out reads containing 'N'");

desc.AddOption("min_length", "", min_length, "minimum length ");

try

{

desc.Parse(argc, argv);

if (argc < 3)

throw logic_error("not enough parameters");

}

catch (exception &e)

{

cerr << e.what() << endl;

cerr << "fq2fa - Convert Fastq sequences to Fasta sequences." << endl;

cerr << "Usage: fq2fa tmp.fq tmp.fa [...] " << endl;

cerr << " fq2fa --paired tmp.fq tmp.fa" << endl;

cerr << " fq2fa --merge tmp_1.fq tmp_2.fq tmp.fa" << endl;

cerr << "Allowed Options: " << endl;

cerr << desc << endl;

exit(1);

}

ReadSequence(argv[1], reads);

cout << "read" << endl;

for (int i = 0; i < (int)reads.size(); i += 2)

{

if (reads[i+1] < reads[i])

swap(reads[i], reads[i+1]);

if ((int)reads[i].size() >= min_length && (int)reads[i+1].size() >= min_length)

aux.push_back(i);

}

sort(aux.begin(), aux.end(), Compare);

cout << "sort" << endl;

int index = 0;

int last = -1;

FastaWriter writer(argv[2]);

for (int i = 0; i < (int)aux.size(); ++i)

{

int id = aux[i];

if (last == -1 || reads[id] != reads[last] || reads[id+1] != reads[last+1])

{

Sequence seq1(reads[id]);

Sequence seq2(reads[id+1]);

writer.Write(seq1, FormatString("reads_%d/1", index));

writer.Write(seq2, FormatString("reads_%d/2", index));

index += 2;

}

last = id;

}

100

101

cout << index << " " << reads.size();

102

103

return 0;

104

}

105

Older »