sim4main.cpp File Reference

#include <stdio.h>
#include <iostream.h>
#include "kzseq.h"
#include <fstream.h>
#include <sys/types.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <fcntl.h>
#include "psublast.h"
#include "sim4.h"
#include "align.h"
#include "sim4b1.h"

Functions
static void	init_stats (sim4_stats_t *)
static void	cds_range (char *, int *, int *)
static void	add_offset_exons (Exon *, int)
static void	add_offset_aligns (edit_script_list *, int)
static void	print_align_blk (const kzseq &so1, const kzseq &so2, edit_script_list **, int, int)
static void	print_align_lat (const kzseq &so1, const kzseq &so2, edit_script_list **, const Exon *, int, int)
void	writeAlign (ostream &os, const exon *&ex, const kzseq &so1, const kzseq &so2, edit_script_list *&als, int ft, int ori, int lengthCut, int percentCut)
int	main (int argc, char *argv[])
Variables
static const char	rcsid []
static const char	Usage []

---

Function Documentation

static void add_offset_aligns	(	edit_script_list *	aligns,
		int	offset
	)			[static]

References edit_script_list::next_script, and edit_script_list::offset2.

Referenced by main().

static void add_offset_exons	(	Exon *	exons,
		int	offset
	)			[static]

References exon::from2, exon::next_exon, exon::to1, and exon::to2.

Referenced by main().

static void cds_range	(	char *	line,
		int *	from,
		int *	to
	)			[static]

References fatal().

Referenced by main().

static void init_stats

(

sim4_stats_t *

st

)

[static]

Referenced by main().

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

*producing the final output ***********/

References add_offset_aligns(), add_offset_exons(), sim4Args::ali_flag, bld_table(), BWD, sim4Args::CDS_from, cds_range(), sim4Args::CDS_to, ckargs(), command, complement_exons(), kzseq::copyRevcomp(), DNA_scores(), EST_GEN, fatal(), file_type, free_align(), free_list(), free_table(), FWD, GEN_EST, sim4Args::getArg(), ifstream(), INIT, init_stats(), LINE, sim4_stats::nmatches, sim4Args::poly_flag, kzseq::polyATsize(), print_align_blk(), print_align_header(), print_align_lat(), print_exons(), print_pipmaker_exons(), readIndex(), kzseq::readSeq(), kzseq::revcomp(), sim4Args::reverse, sim4Args::S, seq1, seq2, kzseq::seq2upper(), kzseq::seqchar(), kzseq::seqhead(), kzseq::seqlen(), SIM4(), Usage, sim4Args::W, and writeAlign().

static void print_align_blk	(	const kzseq &	so1,
		const kzseq &	so2,
		edit_script_list **	Aligns,
		int	file_type,
		int	match_ori
	)			[static]

References align_reverse(), ckalloc(), EST_GEN, Free_script(), FWD, edit_script_list::len1, edit_script_list::len2, edit_script_list::next_script, edit_script_list::offset1, edit_script_list::offset2, print_align(), S, S2A(), edit_script_list::score, edit_script_list::script, seq1, seq2, kzseq::seqchar(), and kzseq::seqlen().

Referenced by main().

static void print_align_lat	(	const kzseq &	so1,
		const kzseq &	so2,
		edit_script_list **	Aligns,
		const Exon *	Exons,
		int	file_type,
		int	match_ori
	)			[static]

References align_reverse(), ckalloc(), EST_GEN, Free_script(), FWD, IDISPLAY(), edit_script_list::len1, edit_script_list::len2, edit_script_list::next_script, edit_script_list::offset1, edit_script_list::offset2, S, S2A(), edit_script_list::script, seq1, seq2, kzseq::seqchar(), and kzseq::seqlen().

Referenced by main(), and writeAlign().

void writeAlign	(	ostream &	os,
		const exon *&	ex,
		const kzseq &	so1,
		const kzseq &	so2,
		edit_script_list *&	als,
		int	ft,
		int	ori,
		int	lengthCut,
		int	percentCut
	)

References BWD, FWD, print_align_lat(), kzseq::seqhead(), and exon::sumLength().

Referenced by main().

---

Variable Documentation

const char rcsid[] [static]

Initial value:

"$Id: sim4main.cpp,v 1.1.1.1 2007-07-13 03:44:06 kzhou Exp $"

const char Usage[] [static]

Initial value:

"%s seq1 seq2_db [[WXKCRDAPNBS]=]\n\n\
       W  -  word size. (W=12)\n\
       X  -  value for terminating word extensions. (X=12)\n\
       K  -  MSP score threshold for the first pass. (e.g., K=16)\n\
       C  -  MSP score threshold for the second pass. (e.g., C=12)\n\
       R  -  direction of search; 0 - search the '+' (direct) strand only; \n\
             1 - search the '-' strand only; 2 - search both strands and \n\
             report the best match. (R=2)\n\
       D  -  bound for the range of diagonals within consecutive msps in an\n\
             exon. (D=10)\n\
       A  -  output format: exon endpoints only (A=0), alignment text (A=1),\n\
             alignment in lav (block) format (A=2), or both exon endpoints\n\
             and alignment text (A=3, A=4). If complement match, A=0,1,2,3\n\
             give direct positions in the long sequence and complement \n\
             positions in the short sequence. A=4 gives direct positions in \n\
             the first sequence, regardless of the relative lengths.\n\
             A=5 prints the exon and CDS coordinates (the latter, if known)\n\
             in the `exon file' format required by PipMaker. To be used\n\
             with full-length mRNA sequences.\n\
       P  -  if not 0, remove poly-A tails; report coordinates in the \n\
             '+' (direct) strand for complement matches; use lav alignment \n\
             headers in all display options. (P=0) \n\
       N  -  accuracy of sequences (non-zero for highly accurate). (N=0)\n\
       B  -  if 0, dis-allow ambiguity codes (other than N and X) in the\n\
             sequence data. (B=1)\n\
       S  -  coding region specification (available only with A=5);\n\
             format: S=n1..n2 \n"

Referenced by main().