params.cpp

unix,linux下编译。用于蛋白质

#include "muscle.h"
#include "objscore.h"
#include "profile.h"
#include "enumopts.h"

const double DEFAULT_MAX_MB_FRACT = 0.8;

SCORE g_scoreCenter = 0;
SCORE g_scoreGapExtend = 0;
SCORE g_scoreGapOpen2 = MINUS_INFINITY;
SCORE g_scoreGapExtend2 = MINUS_INFINITY;
SCORE g_scoreGapAmbig = 0;
SCORE g_scoreAmbigFactor = 0;

extern SCOREMATRIX VTML_LA;
extern SCOREMATRIX PAM200;
extern SCOREMATRIX PAM200NoCenter;
extern SCOREMATRIX VTML_SP;
extern SCOREMATRIX VTML_SPNoCenter;
extern SCOREMATRIX NUC_SP;

PTR_SCOREMATRIX g_ptrScoreMatrix;

const char *g_pstrInFileName = "-";
const char *g_pstrOutFileName = "-";
const char *g_pstrFASTAOutFileName = 0;
const char *g_pstrMSFOutFileName = 0;
const char *g_pstrClwOutFileName = 0;
const char *g_pstrClwStrictOutFileName = 0;
const char *g_pstrHTMLOutFileName = 0;
const char *g_pstrPHYIOutFileName = 0;
const char *g_pstrPHYSOutFileName = 0;
const char *g_pstrDistMxFileName1 = 0;
const char *g_pstrDistMxFileName2 = 0;

const char *g_pstrFileName1 = 0;
const char *g_pstrFileName2 = 0;

const char *g_pstrSPFileName = 0;
const char *g_pstrMatrixFileName = 0;

const char *g_pstrUseTreeFileName = 0;
bool g_bUseTreeNoWarn = false;

const char *g_pstrComputeWeightsFileName;
const char *g_pstrScoreFileName;

const char *g_pstrProf1FileName = 0;
const char *g_pstrProf2FileName = 0;

unsigned g_uSmoothWindowLength = 7;
unsigned g_uAnchorSpacing = 32;
unsigned g_uMaxTreeRefineIters = 1;

unsigned g_uRefineWindow = 200;
unsigned g_uWindowFrom = 0;
unsigned g_uWindowTo = 0;
unsigned g_uSaveWindow = uInsane;
unsigned g_uWindowOffset = 0;

unsigned g_uMaxSubFamCount = 5;

unsigned g_uHydrophobicRunLength = 5;
float g_dHydroFactor = (float) 1.2;

unsigned g_uMinDiagLength = 24;	// TODO alpha -- should depend on alphabet?
unsigned g_uMaxDiagBreak = 1;
unsigned g_uDiagMargin = 5;

float g_dSUEFF = (float) 0.1;

bool g_bPrecompiledCenter = true;
bool g_bNormalizeCounts = false;
bool g_bDiags1 = false;
bool g_bDiags2 = false;
bool g_bAnchors = true;
bool g_bQuiet = false;
bool g_bVerbose = false;
bool g_bRefine = false;
bool g_bRefineW = false;
bool g_bProfDB = false;
bool g_bLow = false;
bool g_bSW = false;
bool g_bClusterOnly = false;
bool g_bProfile = false;
bool g_bPPScore = false;
bool g_bBrenner = false;
bool g_bDimer = false;
bool g_bVersion = false;
bool g_bStable = false;
bool g_bFASTA = false;
bool g_bPAS = false;
bool g_bTomHydro = false;
bool g_bMakeTree = false;

#if	DEBUG
bool g_bCatchExceptions = false;
#else
bool g_bCatchExceptions = true;
#endif

bool g_bMSF = false;
bool g_bAln = false;
bool g_bClwStrict = false;
bool g_bHTML = false;
bool g_bPHYI = false;
bool g_bPHYS = false;

unsigned g_uMaxIters = 8;
unsigned long g_ulMaxSecs = 0;
unsigned g_uMaxMB = 500;

PPSCORE g_PPScore = PPSCORE_LE;
OBJSCORE g_ObjScore = OBJSCORE_SPM;

SEQWEIGHT g_SeqWeight1 = SEQWEIGHT_ClustalW;
SEQWEIGHT g_SeqWeight2 = SEQWEIGHT_ClustalW;

DISTANCE g_Distance1 = DISTANCE_Kmer6_6;
DISTANCE g_Distance2 = DISTANCE_PctIdKimura;

CLUSTER g_Cluster1 = CLUSTER_UPGMB;
CLUSTER g_Cluster2 = CLUSTER_UPGMB;

ROOT g_Root1 = ROOT_Pseudo;
ROOT g_Root2 = ROOT_Pseudo;

bool g_bDiags;

SEQTYPE g_SeqType = SEQTYPE_Auto;

TERMGAPS g_TermGaps = TERMGAPS_Half;

//------------------------------------------------------
// These parameters depending on the chosen prof-prof
// score (g_PPScore), initialized to "Undefined".
float g_dSmoothScoreCeil = fInsane;
float g_dMinBestColScore = fInsane;
float g_dMinSmoothScore = fInsane;
SCORE g_scoreGapOpen = fInsane;
//------------------------------------------------------

static unsigned atou(const char *s)
	{
	return (unsigned) atoi(s);
	}

const char *MaxSecsToStr()
	{
	if (0 == g_ulMaxSecs)
		return "(No limit)";
	return SecsToStr(g_ulMaxSecs);
	}

void ListParams()
	{
	Log("\n");
	Log("%s\n", MUSCLE_LONG_VERSION);
	Log("http://www.drive5.com/muscle\n");
	Log("\n");
	Log("Profile-profile score    %s\n", PPSCOREToStr(g_PPScore));
	Log("Max iterations           %u\n", g_uMaxIters);
	Log("Max trees                %u\n", g_uMaxTreeRefineIters);
	Log("Max time                 %s\n", MaxSecsToStr());
	Log("Max MB                   %u\n", g_uMaxMB);
	Log("Gap open                 %g\n", g_scoreGapOpen);
	Log("Gap extend (dimer)       %g\n", g_scoreGapExtend);
	Log("Gap ambig factor         %g\n", g_scoreAmbigFactor);
	Log("Gap ambig penalty        %g\n", g_scoreGapAmbig);
	Log("Center (LE)              %g\n", g_scoreCenter);
	Log("Term gaps                %s\n", TERMGAPSToStr(g_TermGaps));

	Log("Smooth window length     %u\n", g_uSmoothWindowLength);
	Log("Refine window length     %u\n", g_uRefineWindow);
	Log("Min anchor spacing       %u\n", g_uAnchorSpacing);
	Log("Min diag length (lambda) %u\n", g_uMinDiagLength);
	Log("Diag margin (mu)         %u\n", g_uDiagMargin);
	Log("Min diag break           %u\n", g_uMaxDiagBreak);
	Log("Hydrophobic window       %u\n", g_uHydrophobicRunLength);

	Log("Hydrophobic gap factor   %g\n", g_dHydroFactor);
	Log("Smooth score ceiling     %g\n", g_dSmoothScoreCeil);
	Log("Min best col score       %g\n", g_dMinBestColScore);
	Log("Min anchor score         %g\n", g_dMinSmoothScore);
	Log("SUEFF                    %g\n", g_dSUEFF);

	Log("Brenner root MSA         %s\n", BoolToStr(g_bBrenner));
	Log("Normalize counts         %s\n", BoolToStr(g_bNormalizeCounts));
	Log("Diagonals (1)            %s\n", BoolToStr(g_bDiags1));
	Log("Diagonals (2)            %s\n", BoolToStr(g_bDiags2));
	Log("Anchors                  %s\n", BoolToStr(g_bAnchors));
	Log("MSF output format        %s\n", BoolToStr(g_bMSF));
	Log("Phylip interleaved       %s\n", BoolToStr(g_bPHYI));
	Log("Phylip sequential        %s\n", BoolToStr(g_bPHYS));
	Log("ClustalW output format   %s\n", BoolToStr(g_bAln));
	Log("Catch exceptions         %s\n", BoolToStr(g_bCatchExceptions));
	Log("Quiet                    %s\n", BoolToStr(g_bQuiet));
	Log("Refine                   %s\n", BoolToStr(g_bRefine));
	Log("ProdfDB                  %s\n", BoolToStr(g_bProfDB));
	Log("Low complexity profiles  %s\n", BoolToStr(g_bLow));

	Log("Objective score          %s\n", OBJSCOREToStr(g_ObjScore));

	Log("Distance method (1)      %s\n", DISTANCEToStr(g_Distance1));
	Log("Clustering method (1)    %s\n", CLUSTERToStr(g_Cluster1));
	Log("Root method (1)          %s\n", ROOTToStr(g_Root1));
	Log("Sequence weighting (1)   %s\n", SEQWEIGHTToStr(g_SeqWeight1));

	Log("Distance method (2)      %s\n", DISTANCEToStr(g_Distance2));
	Log("Clustering method (2)    %s\n", CLUSTERToStr(g_Cluster2));
	Log("Root method (2)          %s\n", ROOTToStr(g_Root2));
	Log("Sequence weighting (2)   %s\n", SEQWEIGHTToStr(g_SeqWeight2));

	Log("\n");
	}

static void SetDefaultsLE()
	{
	g_ptrScoreMatrix = &VTML_LA;

	//g_scoreGapOpen = (SCORE) -3.00;
	//g_scoreCenter = (SCORE) -0.55;
	g_scoreGapOpen = (SCORE) -2.9;
	g_scoreCenter = (SCORE) -0.52;

	g_bNormalizeCounts = true;

	//g_dSmoothScoreCeil = 5.0;
	//g_dMinBestColScore = 4.0;
	//g_dMinSmoothScore = 2.0;
	g_dSmoothScoreCeil = 3.0;
	g_dMinBestColScore = 2.0;
	g_dMinSmoothScore = 1.0;

	g_Distance1 = DISTANCE_Kmer6_6;
	g_Distance2 = DISTANCE_PctIdKimura;
	}

static void SetDefaultsSP()
	{
	g_ptrScoreMatrix = &PAM200;

	g_scoreGapOpen = -1439;
	g_scoreCenter = 0.0;	// center pre-added into score mx

	g_bNormalizeCounts = false;

	g_dSmoothScoreCeil = 200.0;
	g_dMinBestColScore = 300.0;
	g_dMinSmoothScore = 125.0;

	g_Distance1 = DISTANCE_Kmer6_6;
	g_Distance2 = DISTANCE_PctIdKimura;
	}

static void SetDefaultsSV()
	{
	g_ptrScoreMatrix = &VTML_SP;

	g_scoreGapOpen = -300;
	g_scoreCenter = 0.0;	// center pre-added into score mx

	g_bNormalizeCounts = false;

	g_dSmoothScoreCeil = 90.0;
	g_dMinBestColScore = 130.0;
	g_dMinSmoothScore = 40.0;

	g_Distance1 = DISTANCE_Kmer6_6;
	g_Distance2 = DISTANCE_PctIdKimura;
	}

//static void SetDefaultsSPN()
//	{
//	g_ptrScoreMatrix = &NUC_SP;
//
//	g_scoreGapOpen = -400;
//	g_scoreCenter = 0.0;	// center pre-added into score mx
//
//	g_bNormalizeCounts = false;
//
//	g_dSmoothScoreCeil = 999.0;		// disable
//	g_dMinBestColScore = 90;
//	g_dMinSmoothScore = 90;
//
//	g_Distance1 = DISTANCE_Kmer4_6;
//	g_Distance2 = DISTANCE_PctIdKimura;
//	}

static void SetDefaultsSPN_DNA()
	{
	g_ptrScoreMatrix = &NUC_SP;

	g_scoreGapOpen = -400;
	g_scoreCenter = 0.0;	// center pre-added into score mx
	g_scoreGapExtend = 0.0;

	g_bNormalizeCounts = false;

	g_dSmoothScoreCeil = 999.0;		// disable
	g_dMinBestColScore = 90;
	g_dMinSmoothScore = 90;

	g_Distance1 = DISTANCE_Kmer4_6;
	g_Distance2 = DISTANCE_PctIdKimura;
	}

static void SetDefaultsSPN_RNA()
	{
	g_ptrScoreMatrix = &NUC_SP;

	g_scoreGapOpen = -420;
	g_scoreCenter = -300;	// total center = NUC_EXTEND - 300 
	g_scoreGapExtend = 0.0;