int main()
{
// push waypoints to a routes' map
waypoints.push_back(waypoint(-4.0, -3.0)); // 0
waypoints.push_back(waypoint(-3.0, -1.0)); // 1
waypoints.push_back(waypoint(1.0, -4.0)); // 2
waypoints.push_back(waypoint(3.0, -2.0)); // 3
waypoints.push_back(waypoint(-1.0, 1.0)); // 4
waypoints.push_back(waypoint(4.0, -3.0)); // 5
waypoints.push_back(waypoint(2.0, 1.0)); // 6
waypoints.push_back(waypoint(-2.0, 2.0)); // 7
waypoints.push_back(waypoint(5.0, -1.0)); // 8
waypoints.push_back(waypoint(6.0, 2.0)); // 9
waypoints.push_back(waypoint(-1.0, 4.0)); // 10
waypoints.push_back(waypoint(4.0, 4.0)); // 11
waypoints.push_back(waypoint(8.0, 3.0)); // 12
waypoints.push_back(waypoint(1.0, 6.0)); // 13
waypoints.push_back(waypoint(6.0, 5.0)); // 14
waypoints.push_back(waypoint(2.0, 7.0)); // 15
waypoints.push_back(waypoint(7.0, 7.0)); // 16
waypoints.push_back(waypoint(4.0, 7.0)); // 17
waypoints.push_back(waypoint(6.0, 9.0)); // 18
waypoints.push_back(waypoint(8.0, 9.0)); // 19
obstacles.push_back(obstacle(0.0,5.0,1.0)); // 0
obstacles.push_back(obstacle(3.0,3.0,2.0)); // 1
obstacles.push_back(obstacle(3.5,0.0,1.0)); // 2
obstacles.push_back(obstacle(8.0,7.0,2.5)); // 3
// define exponential traverser type and create an exemplar
typedef ExpTr<node_value,cost_value,expand,less<cost_value> > ExpTr_t;
ExpTr_t j;
cout << "in exp traverser forward search ...\n";
j = ExpTr_t(0);
// set a maximal lag length
j.getexpfunc().setMaxLagLen(5.0); // or 6.0
// add two defined handlers to enable calculation of an expansions number
j.set_handler_on_expand_root(OnExpandRoot);
j.set_handler_on_select_cursor(OnSelectCursor);
// launch A* algorithm (combination of a predefined heuristics with a best - first
// search that is provided by exponential traverser is A*!)
ForwardSearch(j, goal);
ShowPath(j);
// show founded path
cout << "number of expansions = " << num_expansions << '\n';
cout << "...out \n\n";
cout << "Search is completed.\n";
return 0;
}
void ForwardSearch( int u, int inDummy, int time, int *visitTime, int *counter, int *visitDummy, ContigGraph &contigGraph )
{
if ( visitTime[u] == 2 * time )
return ;
visitTime[u] = 2 * time ;
counter[u] = 1 ;
visitDummy[u] = inDummy ;
struct _pair *buffer = new struct _pair[ MAX_NEIGHBOR ] ;
int ncnt ;
int i ;
ncnt = contigGraph.GetNeighbors( u, 1 - inDummy, buffer, MAX_NEIGHBOR ) ;
for ( i = 0 ; i < ncnt ; ++i )
{
/*if ( time == 639 )
{
printf( "forwardsearch: (%d %d)=>(%d %d)\n", u, 1 - inDummy, buffer[i].a, buffer[i].b ) ;
}*/
ForwardSearch( buffer[i].a, buffer[i].b, time, visitTime, counter, visitDummy, contigGraph ) ;
}
delete[] buffer ;
}
int main()
{
// grid map
// vertical edges
C[0][1] = 11;
C[1][2] = 10;
C[2][3] = 8;
C[3][4] = 7;
C[4][5] = 9;
C[5][6] = 10;
C[10][11] = 9;
C[11][12] = 8;
C[12][13] = 7;
C[13][14] = 6;
C[14][15] = 8;
C[15][16] = 12;
C[20][21] = 10;
C[21][22] = 10;
C[22][23] = 8;
C[23][24] = 8;
C[24][25] = 9;
C[25][26] = 13;
C[30][31] = 12;
C[31][32] = 11;
C[32][33] = 9;
C[33][34] = 7;
C[34][35] = 10;
C[35][36] = 13;
C[40][41] = 13;
C[41][42] = 10;
C[42][43] = 9;
C[43][44] = 8;
C[44][45] = 11;
C[45][46] = 14;
C[50][51] = 14;
C[51][52] = 11;
C[52][53] = 10;
C[53][54] = 10;
C[54][55] = 12;
C[55][56] = 15;
C[60][61] = 14;
C[61][62] = 12;
C[62][63] = 11;
C[63][64] = 11;
C[64][65] = 13;
C[65][66] = 14;
C[70][71] = 12;
C[71][72] = 10;
C[72][73] = 8;
C[73][74] = 10;
C[74][75] = 12;
C[75][76] = 13;
C[80][81] = 10;
C[81][82] = 9;
C[82][83] = 7;
C[83][84] = 9;
C[84][85] = 12;
C[85][86] = 14;
// horizontal edges
C[0][10] = 12;
C[10][20] = 11;
C[20][30] = 10;
C[30][40] = 9;
C[40][50] = 13;
C[50][60] = 14;
C[60][70] = 17;
C[70][80] = 20;
C[1][11] = 11;
C[11][21] = 9;
C[21][31] = 8;
C[31][41] = 7;
C[41][51] = 9;
C[51][61] = 13;
C[61][71] = 14;
C[71][81] = 19;
C[2][12] = 10;
C[12][22] = 8;
C[22][32] = 8;
C[32][42] = 8;
C[42][52] = 10;
C[52][62] = 12;
C[62][72] = 13;
C[72][82] = 15;
C[3][13] = 12;
C[13][23] = 13;
C[23][33] = 12;
C[33][43] = 10;
C[43][53] = 11;
C[53][63] = 13;
C[63][73] = 15;
C[73][83] = 20;
C[4][14] = 14;
C[14][24] = 13;
C[24][34] = 12;
C[34][44] = 10;
C[44][54] = 9;
C[54][64] = 8;
C[64][74] = 11;
C[74][84] = 15;
C[5][15] = 17;
C[15][25] = 14;
C[25][35] = 13;
C[35][45] = 10;
C[45][55] = 11;
C[55][65] = 13;
C[65][75] = 14;
C[75][85] = 17;
C[6][16] = 20;
C[16][26] = 18;
C[26][36] = 16;
C[36][46] = 15;
C[46][56] = 14;
C[56][66] = 12;
C[66][76] = 15;
C[76][86] = 17;
// diagonal edges
C[5][16] = 28;
C[4][15] = 23;
C[15][26] = 20;
C[3][14] = 18;
C[14][25] = 21;
C[25][36] = 27;
C[2][13] = 19;
C[13][24] = 20;
C[24][35] = 21;
C[35][46] = 25;
C[1][12] = 18;
C[12][23] = 17;
C[23][34] = 21;
C[34][45] = 19;
C[45][56] = 27;
C[0][11] = 20;
C[11][22] = 18;
C[22][33] = 18;
C[33][44] = 19;
C[44][55] = 22;
C[55][66] = 26;
C[10][21] = 20;
C[21][32] = 19;
C[32][43] = 18;
C[43][54] = 20;
C[54][65] = 22;
C[65][76] = 28;
C[20][31] = 20;
C[31][42] = 18;
C[42][53] = 19;
C[53][64] = 21;
C[64][75] = 22;
C[75][86] = 29;
C[30][41] = 20;
C[41][52] = 19;
C[52][63] = 22;
C[63][74] = 23;
C[74][85] = 25;
C[40][51] = 23;
C[51][62] = 24;
C[62][73] = 22;
C[73][84] = 24;
C[50][61] = 24;
C[61][72] = 23;
C[72][83] = 24;
C[60][71] = 26;
C[71][82] = 26;
C[70][81] = 29;
typedef LinTr<int,int,expand,greater<int> > LinTr_t; // linear ttraverser
typedef ExpTr<int,int,expand,less<int> > ExpTr_t; // exponential traverser
typedef ExpTr<int,int,bidir_expand,less<int> > RevExpTr_t; // second traverser for bi-directional search
LinTr_t i;
ExpTr_t j;
RevExpTr_t k;
// LINEAR TRAVERSER IN WORK
/*
cout << "in lin traverser forward search ...\n";
cout << "eps = " << eps << '\n';
i = LinTr_t(start);
i.set_handler_on_expand_root(OnExpandRoot);
i.set_handler_on_select_cursor(OnSelectCursor);
ForwardSearch(i, goal);
ShowPath(i);
cout << "number of expansions = " << num_expansions << '\n';
cout << "...out \n\n";
/**/
/*
cout << "in lin traverser search bounded by " << thresh << "...\n";
BoundedSearch(i = LinTr_t(start), goal, thresh);
ShowPath(i);
cout << "...out \n\n";
cout << "in lin traverser search bounded by " << next_thresh << "... continue previous one ... \n";
BoundedSearch(i, goal, next_thresh);
ShowPath(i);
cout << "...out \n\n";
cout << "in lin traverser search bounded by " << next_thresh2 << "... continue previous one ...\n";
BoundedSearch(i, goal, next_thresh2);
ShowPath(i);
cout << "...out \n\n";
*/
/*
cout << "in lin traverser search bounded by " << less_than_opt_value << "...\n";
BoundedSearch(i = LinTr_t(start), goal, less_than_opt_value);
ShowPath(i);
cout << "...out \n\n";
*/
/*
cout << "in lin traverser optimal search bounded by " << thresh << " from initial state ...\n";
i = LinTr_t(start);
BoundedOptSearch(i, goal, thresh);
ShowPath(i);
cout << "...out\n\n";
*/
/*
cout << "in lin traverser optimal search bounded by " << opt_value << " from initial state ...\n";
BoundedOptSearch(i = LinTr_t(start), goal, opt_value);
ShowPath(i);
cout << "...out\n\n";
*/
/**/
cout << "in lin traverser branch and bound ...\n";
i = LinTr_t(start);
int eps = i.getexpfunc().GetEps();
cout << "eps = " << eps << '\n';
i.set_handler_on_expand_root(OnExpandRoot);
i.set_handler_on_select_cursor(OnSelectCursor);
BranchAndBound(i, goal);
ShowPath(i);
cout << "number of expansions = " << num_expansions << '\n';
cout << "...out \n\n";
cout << "in lin traverser branch and bound ...\n";
i = LinTr_t(start);
i.getexpfunc().SetEps(10);
expand &e = i.getexpfunc();
e.SetEps(12);
eps = i.getexpfunc().GetEps();
cout << "eps = " << eps << "\n";
i.set_handler_on_expand_root(OnExpandRoot);
i.set_handler_on_select_cursor(OnSelectCursor);
BranchAndBound(i, goal);
ShowPath(i);
cout << "number of expansions = " << num_expansions << '\n';
cout << "...out\n\n";
cout << "in lin traverser search bounded by " << 140 << "...\n";
cout << "eps = " << eps << '\n';
i = LinTr_t(start);
i.set_handler_on_expand_root(OnExpandRoot);
i.set_handler_on_select_cursor(OnSelectCursor);
BoundedSearch(i, goal, 140);
ShowPath(i);
cout << "number of expansions = " << num_expansions << '\n';
cout << "...out\n\n";
/**/
// EXPONENTIAL TRAVERSER IN WORK
/**/
cout << "in exp traverser forward search ...\n";
j = ExpTr_t(start);
j.getexpfunc().SetEps(0);
eps = j.getexpfunc().GetEps();
cout << "eps = " << eps << '\n';
j.set_handler_on_expand_root(OnExpandRoot);
j.set_handler_on_select_cursor(OnSelectCursor);
ForwardSearch(j, goal);
ShowPath(j);
cout << "number of expansions = " << num_expansions << '\n';
cout << "...out \n\n";
/**/
cout << "in exp traverser search bounded by " << thresh << "...\n";
BoundedSearch(j = ExpTr_t(start), goal, thresh);
ShowPath(j);
cout << "...out \n\n";
/**/
/*
cout << "in exp traverser search bounded by " << next_thresh << "... continue previous one...\n";
BoundedSearch(j, goal, next_thresh);
ShowPath(j);
cout << "...out \n\n";
/**/
/*
cout << "in exp traverser optimal search from initial state bounded by " << thresh << "...\n";
BoundedOptSearch(j = ExpTr_t(start), goal, thresh);
ShowPath(j);
cout << "...out \n\n";
/**/
/*
cout << "in exp traverser branch and bound ...\n";
BranchAndBound(j = ExpTr_t(start), goal);
ShowPath(j);
cout << "...out \n\n";
*/
/**/
cout << "in exp traverser iterative deepening ...\n";
j = ExpTr_t(start);
j.getexpfunc().SetEps(10);
eps = j.getexpfunc().GetEps();
cout << "eps = " << eps << '\n';
j.set_handler_on_expand_root(OnExpandRoot);
j.set_handler_on_select_cursor(OnSelectCursor);
IterativeDeepening(j, goal);
ShowPath(j);
cout << "number of expansions = " << num_expansions << '\n';
cout << "...out \n\n";
/**/
/*
cout << "in exp traverser bi-directional search ...\n";
BidirectionalSearch( j = ExpTr_t(start), k = RevExpTr_t(final) );
ShowStickedPaths(j,k);
cout << "...out \n\n";
*/
cout << "All searches completed.\n";
return 0;
}
int main( int argc, char *argv[] )
{
Alignments alignments ;
Genome genome ;
std::vector<int> rascafFileId ;
char line[2048] ;
char prefix[512] = "rascaf_scaffold" ;
int rawAssemblyInd = 1 ;
FILE *rascafFile ;
bool contigLevel = false ;
int i ;
FILE *outputFile ;
FILE *infoFile ;
breakN = 1 ;
if ( argc < 2 )
{
fprintf( stderr, "%s", usage ) ;
exit( 1 ) ;
}
for ( i = 1 ; i < argc ; ++i )
{
if ( !strcmp( "-o", argv[i] ) )
{
strcpy( prefix, argv[i + 1 ] ) ;
++i ;
}
else if ( !strcmp( "-ms", argv[i] ) )
{
minSupport = atoi( argv[i + 1] ) ;
++i ;
}
else if ( !strcmp( "-ignoreGap", argv[i] ) )
{
ignoreGap = true ;
}
else if ( !strcmp( "-r", argv[i] ) )
{
rascafFileId.push_back( i + 1 ) ;
++i ;
}
else
{
fprintf( stderr, "Unknown option: %s\n", argv[i] ) ;
exit( EXIT_FAILURE ) ;
}
}
if ( rascafFileId.size() == 0 )
{
fprintf( stderr, "Must use -r to specify rascaf output file.\n" ) ;
exit( EXIT_FAILURE ) ;
}
MAX_NEIGHBOR = 1 + rascafFileId.size() ;
// Get the bam file.
rascafFile = fopen( argv[ rascafFileId[0] ], "r" ) ;
while ( fgets( line, sizeof( line ), rascafFile ) != NULL )
{
if ( strstr( line, "command line:" ) )
{
char *p ;
char buffer[512] ;
p = strstr( line, "-breakN" ) ;
if ( p != NULL )
{
p += 7 ;
while ( *p == ' ' )
++p ;
for ( i = 0 ; *p && *p != ' ' ; ++p, ++i )
buffer[i] = *p ;
buffer[i] = '\0' ;
breakN = atoi( buffer ) ;
}
p = strstr( line, "-b" ) ;
if ( p == NULL )
{
fprintf( stderr, "Could not find the bam file specified by -b in Rascaf.\n" ) ;
exit( 1 ) ;
}
p += 2 ;
while ( *p == ' ' )
++p ;
for ( i = 0 ; *p && *p != ' ' ; ++p, ++i )
buffer[i] = *p ;
buffer[i] = '\0' ;
alignments.Open( buffer ) ;
p = strstr( line, "-f") ;
if ( p == NULL )
{
fprintf( stderr, "Could not find the raw assembly file specified by -f in Rascaf.\n" ) ;
exit( 1 ) ;
}
p += 2 ;
while ( *p == ' ' )
++p ;
for ( i = 0 ; *p && *p != ' ' && *p != '\n' ; ++p, ++i )
buffer[i] = *p ;
buffer[i] = '\0' ;
fprintf( stderr, "Found raw assembly file: %s\n", buffer ) ;
genome.Open( alignments, buffer ) ;
break ;
}
}
fclose( rascafFile ) ;
// Parse the input.
for ( unsigned int fid = 0 ; fid < rascafFileId.size() ; ++fid )
{
rascafFile = fopen( argv[ rascafFileId[fid] ], "r" ) ;
bool start = false ;
int tag ;
while ( fgets( line, sizeof( line ), rascafFile ) != NULL )
{
if ( strstr( line, "command line:" ) )
{
start = true ;
if ( strstr( line, "-f" ) )
{
contigLevel = true ;
}
continue ;
}
if ( !start )
continue ;
if ( !strcmp( line, "WARNINGS:\n" ) )
break ;
std::vector<struct _part> nparts ;
if ( line[0] >= '0' && line[0] <= '9' )
{
AddConnection( line, alignments, nparts ) ;
connects.push_back( nparts ) ;
tag = 0 ;
}
else if ( line[0] == '\t' || line[0] == ' ' )
{
// Break the nparts if the support is too low.
int num = 0 ;
for ( i = 0 ; line[i] < '0' || line[i] > '9' ; ++i )
;
for ( ; line[i] >= '0' && line[i] <= '9' ; ++i )
num = num * 10 + line[i] - '0' ;
++tag ;
if ( num < minSupport )
{
nparts = connects.back() ;
connects.pop_back() ;
int size = nparts.size() ;
std::vector<struct _part> newNParts ;
for ( i = 0 ; i < tag ; ++i )
newNParts.push_back( nparts[i] ) ;
if ( newNParts.size() > 1 )
connects.push_back( newNParts ) ;
newNParts.clear() ;
for ( ; i < size ; ++i )
newNParts.push_back( nparts[i] ) ;
if ( newNParts.size() > 1 )
connects.push_back( newNParts ) ;
tag = 0 ;
}
}
}
fclose( rascafFile ) ;
}
if ( contigLevel == false )
{
genome.SetIsOpen( contigLevel ) ;
}
// Build the graph
int contigCnt = genome.GetContigCount() ;
int edgeCnt = 0 ;
int csize = connects.size() ;
for ( i = 0 ; i < csize ; ++i )
edgeCnt += connects[i].size() ;
ContigGraph contigGraph( contigCnt, contigCnt + edgeCnt ) ;
for ( i = 0 ; i < contigCnt - 1 ; ++i )
{
if ( genome.GetChrIdFromContigId( i ) == genome.GetChrIdFromContigId( i + 1 ) )
{
contigGraph.AddEdge( i, 1, i + 1, 0 ) ;
}
}
for ( i = 0 ; i < csize ; ++i )
{
std::vector<struct _part> &parts = connects[i] ;
int size = parts.size() ;
for ( int j = 0 ; j < size - 1 ; ++j )
{
struct _part &a = parts[j] ;
struct _part &b = parts[j + 1] ;
// Two dummy nodes for each contig. Left is 0, right is 1
int dummyU = 0 ;
int dummyV = 0 ;
if ( a.strand == '+' )
dummyU = 1 ;
if ( b.strand == '-' )
dummyV = 1 ;
contigGraph.AddEdge( a.contigId, dummyU, b.contigId, dummyV, true ) ;
}
}
// Check the cycles in the contig graph. This may introduces when combining different rascaf outputs.
int *visitTime = new int[contigCnt] ;
struct _pair *neighbors = new struct _pair[ MAX_NEIGHBOR ] ;
bool *isInCycle = new bool[contigCnt] ;
std::vector<int> cycleNodes ;
memset( visitTime, -1, sizeof( int ) * contigCnt ) ;
memset( isInCycle, false, sizeof( bool ) * contigCnt ) ;
for ( i = 0 ; i < contigCnt ; ++i )
{
if ( isInCycle[i] )
continue ;
if ( contigGraph.IsInCycle( i, cycleNodes, visitTime ) )
{
int cnt = cycleNodes.size() ;
//printf( "===\n") ;
for ( int j = 0 ; j < cnt ; ++j )
{
//printf( "In cycle %d\n", cycleNodes[j] ) ;
isInCycle[ cycleNodes[j] ] = true ;
}
}
}
//exit( 1 ) ;
// Remove the connected edges involving the nodes in the cycle
for ( i = 0 ; i < contigCnt ; ++i )
{
if ( isInCycle[i] )
{
for ( int dummy = 0 ; dummy <= 1 ; ++dummy )
{
int ncnt = contigGraph.GetNeighbors( i, dummy, neighbors, MAX_NEIGHBOR ) ;
for ( int j = 0 ; j < ncnt ; ++j )
{
if ( neighbors[j].a == i + 2 * dummy - 1 && neighbors[j].b != dummy
&& genome.GetChrIdFromContigId( i ) == genome.GetChrIdFromContigId( neighbors[j].a ) )
continue ; // the connection created by the raw assembly
else
contigGraph.RemoveEdge( i, dummy, neighbors[j].a, neighbors[j].b ) ;
}
}
}
}
//delete[] isInCycle ;
//printf( "hi: %d %d\n", __LINE__, contigCnt ) ;
//printf( "%d %d\n", contigGraph.GetNeighbors( 0, 0, neighbors, MAX_NEIGHBOR ), contigGraph.GetNeighbors( 0, 1, neighbors, MAX_NEIGHBOR ) ) ;
// Sort the scaffolds from fasta file, so that longer scaffold come first
int scafCnt = genome.GetChrCount() ;
struct _pair *scafInfo = new struct _pair[scafCnt] ;
memset( scafInfo, -1, sizeof( struct _pair) * scafCnt ) ;
for ( i = 0 ; i < contigCnt ; ++i )
{
int chrId = genome.GetChrIdFromContigId( i ) ;
if ( scafInfo[chrId].a == -1 )
{
scafInfo[ chrId ].a = i ;
scafInfo[ chrId ].b = genome.GetChrLength( chrId ) ;
}
}
qsort( scafInfo, scafCnt, sizeof( struct _pair ), CompScaffold ) ;
// Merge the branches and build the scaffold
ContigGraph scaffold( contigCnt, 2 * contigCnt ) ;
// Use a method similar to topological sort
bool *used = new bool[contigCnt] ;
int *degree = new int[2 *contigCnt] ;
int *danglingVisitTime = new int[contigCnt] ;
int *counter = new int[contigCnt] ;
int *visitDummy = new int[ contigCnt ] ;
int *buffer = new int[contigCnt] ;
int *buffer2 = new int[contigCnt] ;
bool *isInQueue = new bool[ contigCnt ] ;
int *chosen = new int[contigCnt] ;
int chosenCnt ;
memset( isInCycle, false, sizeof( bool ) * contigCnt ) ;
memset( visitTime, -1, sizeof( int ) * contigCnt ) ;
memset( visitDummy, -1, sizeof( int ) * contigCnt ) ;
memset( counter, -1, sizeof( int ) * contigCnt ) ;
// Use those memory to remove triangular cycles
for ( i = 0 ; i < scafCnt ; ++i )
{
int from, to ;
if ( scafInfo[i].a == -1 )
continue ;
genome.GetChrContigRange( genome.GetChrIdFromContigId( scafInfo[i].a ), from, to ) ;
ForwardSearch( from, 0, i, visitTime, counter, visitDummy, contigGraph ) ;
chosenCnt = 0 ;
BackwardSearchForTriangularCycle( to, 1, i, visitTime, counter, visitDummy, contigGraph, chosen, chosenCnt ) ;
for ( int j = 0 ; j < chosenCnt ; ++j )
{
//printf( "%d\n", chosen[j] ) ;
isInCycle[ chosen[j] ] = true ;
}
}
for ( i = 0 ; i < contigCnt ; ++i )
{
if ( isInCycle[i] )
{
for ( int dummy = 0 ; dummy <= 1 ; ++dummy )
{
int ncnt = contigGraph.GetNeighbors( i, dummy, neighbors, MAX_NEIGHBOR ) ;
for ( int j = 0 ; j < ncnt ; ++j )
{
if ( neighbors[j].a == i + 2 * dummy - 1 && neighbors[j].b != dummy
&& genome.GetChrIdFromContigId( i ) == genome.GetChrIdFromContigId( neighbors[j].a ) )
continue ; // the connection created by the raw assembly
else
contigGraph.RemoveEdge( i, dummy, neighbors[j].a, neighbors[j].b ) ;
}
}
}
}
memset( used, false, sizeof( bool ) * contigCnt ) ;
memset( visitTime, -1, sizeof( int ) * contigCnt ) ;
memset( visitDummy, -1, sizeof( int ) * contigCnt ) ;
memset( danglingVisitTime, -1, sizeof( int ) * contigCnt ) ;
memset( counter, -1, sizeof( int ) * contigCnt ) ;
memset( isInQueue, false, sizeof( bool ) * contigCnt ) ;
ContigGraph newGraph( contigCnt, edgeCnt ) ;
// Compute the gap size
int *gapSize = new int[contigCnt] ;
for ( i = 0 ; i < contigCnt - 1 ; ++i )
{
if ( genome.GetChrIdFromContigId( i ) == genome.GetChrIdFromContigId( i + 1 ) )
{
struct _contig c1 = genome.GetContigInfo( i ) ;
struct _contig c2 = genome.GetContigInfo( i + 1 ) ;
gapSize[i] = c2.start - c1.end - 1 ;
}
else
gapSize[i] = -1 ;
}
// Start search
int ncnt ;
struct _pair *queue = new struct _pair[ contigCnt ] ;
int head = 0, tail ;
int danglingTime = 0 ;
// Pre-allocate the subgraph.
ContigGraph subgraph( contigCnt, 3 * contigCnt ) ;
for ( i = 0 ; i < scafCnt ; ++i )
{
//if ( used[144281] == true )
// printf( "changed %d %d\n", i, scafInfo[i - 1].a ) ;
if ( scafInfo[i].a == -1 )
continue ;
int from, to ;
genome.GetChrContigRange( genome.GetChrIdFromContigId( scafInfo[i].a ), from, to ) ;
//printf( "%d: %d %d %d\n", i, scafInfo[i].b, from, to ) ;
ForwardSearch( from, 0, i, visitTime, counter, visitDummy, contigGraph ) ;
chosenCnt = 0 ;
BackwardSearch( to, 1, i, visitTime, counter, contigGraph, chosen, chosenCnt ) ;
/*printf( "%s %d (%d %d) %d\n", alignments.GetChromName( genome.GetChrIdFromContigId( scafInfo[i].a ) ), i, from, to, chosenCnt ) ;
if ( chosenCnt > 1 )
{
printf( "=== " ) ;
for ( int j = 0 ; j < chosenCnt ; ++j )
printf( "%d ", chosen[j] ) ;
printf( "\n" ) ;
}*/
for ( int j = 0 ; j < chosenCnt ; ++j )
{
ncnt = contigGraph.GetNeighbors( chosen[j], 0, neighbors, MAX_NEIGHBOR ) ;
//printf( "%d %d %d: %d %d %d\n", j, chosen[j], ncnt, neighbors[0].a, visitTime[ neighbors[0].a ],
// counter[neighbors[0].a ] ) ;
for ( int k = 0 ; k < ncnt ; ++k )
{
//if ( i == 639 )
// printf( "Neighbor from 0 %d: %d %d\n", k, neighbors[k].a, neighbors[k].b ) ;
if ( visitTime[ neighbors[k].a ] == 2 * i + 1 && counter[neighbors[k].a ] == 2 )
{
subgraph.AddEdge( chosen[j], 0, neighbors[k].a, neighbors[k].b, true ) ;
//printf( "subgraph: (%d %d)=>(%d %d)\n", chosen[j], 0, neighbors[k].a, neighbors[k].b ) ;
}
}
ncnt = contigGraph.GetNeighbors( chosen[j], 1, neighbors, MAX_NEIGHBOR ) ;
for ( int k = 0 ; k < ncnt ; ++k )
{
//if ( i == 639 )
// printf( "Neighbor from 1 %d: %d %d\n", k, neighbors[k].a, neighbors[k].b ) ;
if ( visitTime[ neighbors[k].a ] == 2 * i + 1 && counter[neighbors[k].a ] == 2 )
{
subgraph.AddEdge( chosen[j], 1, neighbors[k].a, neighbors[k].b, true ) ;
//printf( "subgraph: (%d %d)=>(%d %d)\n", chosen[j], 1, neighbors[k].a, neighbors[k].b ) ;
}
}
}
// Initialize the degree counter
for ( int j = 0 ; j < chosenCnt ; ++j )
{
for ( int l = 0 ; l < 2 ; ++l )
{
/*if ( i == 6145 )
{
std::vector<struct _pair> neighbors ;
ncnt = subgraph.GetNeighbors( chosen[j], l, neighbors ) ;
printf( "%d ncnt=%d\n", l, ncnt ) ;
}*/
ncnt = subgraph.GetNeighbors( chosen[j], l, neighbors, MAX_NEIGHBOR ) ;
degree[ 2 * chosen[j] + l ] = ncnt ;
}
}
// "topological" sort
head = 0 ;
isInQueue[from] = true ;
queue[0].a = from ;
queue[0].b = 0 ;
tail = 1 ;
int prevTag = -1 ;
int *prevAdd = buffer ; // reuse counter to save some memory.
int *nextAdd = buffer2 ;
int firstAdd = -1 ;
while ( head < tail )
{
int tailTag = tail ;
for ( int j = head ; j < tailTag ; ++j )
{
nextAdd[j] = -1 ;
if ( !used[ queue[j].a ] )
{
used[ queue[j].a ] = true ;
if ( prevTag != -1 )
{
scaffold.AddEdge( queue[ prevTag].a, 1 - queue[prevTag].b, queue[j].a, queue[j].b ) ;
nextAdd[ prevTag ] = j ;
/*if ( i == 639 )
printf( "(%lld %lld)=>(%lld %lld)\n", queue[ prevTag].a, 1 - queue[prevTag].b, queue[j].a, queue[j].b ) ;*/
}
else
firstAdd = j ;
prevTag = j ;
}
prevAdd[j] = prevTag ; // the most recent(<=) queue id when added to scaffold.
ncnt = subgraph.GetNeighbors( queue[j].a, 1 - queue[j].b, neighbors, MAX_NEIGHBOR ) ;
for ( int k = 0 ; k < ncnt ; ++k )
{
--degree[ 2 * neighbors[k].a + neighbors[k].b ] ;
if ( degree[ 2 * neighbors[k].a + neighbors[k].b ] == 0 && !isInQueue[neighbors[k].a] )
{
isInQueue[ neighbors[k].a ] = true ;
queue[ tail ] = neighbors[k] ; // Interesting assignment, I think.
++tail ;
/*if ( i == 639 )
printf( "pushed in queue: %d\n", neighbors[k].a ) ;*/
// Put the consecutive contigs together.
struct _pair testNeighbors[ MAX_NEIGHBOR ] ;
struct _pair tag ;
tag = neighbors[k] ;
while ( 1 )
{
if ( contigGraph.GetNeighbors( tag.a, 1 - tag.b, testNeighbors, MAX_NEIGHBOR ) != 1 )
break ;
int n = subgraph.GetNeighbors( tag.a, 1 - tag.b, testNeighbors, MAX_NEIGHBOR ) ;
if ( n != 1 )
break ;
//printf( "%d %d\n", n, testNeighbors[0].a ) ;
struct _pair backNeighbors[ MAX_NEIGHBOR ] ;
if ( contigGraph.GetNeighbors( testNeighbors[0].a, testNeighbors[0].b, backNeighbors, MAX_NEIGHBOR ) != 1 )
break ;
n = subgraph.GetNeighbors( testNeighbors[0].a, testNeighbors[0].b, backNeighbors, MAX_NEIGHBOR ) ;
if ( n != 1 )
break ;
isInQueue[ testNeighbors[0].a ] = true ;
queue[tail] = testNeighbors[0] ;
++tail ;
/*if ( i == 639 )
printf( "pushed in queue: %d\n", testNeighbors[0].a ) ;*/
tag = testNeighbors[0] ;
}
}
}
}
head = tailTag ;
}
// Remove the effect on the subgraph.
/*if ( tail != chosenCnt )
{
printf( "WARNING: not matched\n" ) ;
exit( 1 ) ;
}*/
for ( int j = 0 ; j < tail ; ++j )
{
visitDummy[ queue[j].a ] = -1 ;
counter[ queue[j].a ] = -1 ;
subgraph.RemoveAdjacentEdges( queue[j].a ) ;
isInQueue[ queue[j].a ] = false ;
}
subgraph.ResetEdgeUsed() ;
// no point is picked
if ( prevTag == -1 )
{
continue ;
}
// Update the gap size
prevTag = -1 ;
for ( int j = 0 ; j < tail - 1 ; ++j )
{
if ( genome.GetChrIdFromContigId( queue[j].a ) == genome.GetChrIdFromContigId( from ) )
prevTag = queue[j].a ;
else if ( prevTag != -1 )
{
struct _contig c = genome.GetContigInfo( queue[j].a ) ;
gapSize[prevTag] -= ( c.end - c.start + 1) ;
}
}
// Add the dangling contigs. Use the fact that the queue holding the contigs in the same order as in the scaffold.
// 5'->3' dangling
int *chosenDummy = degree ;
for ( int j = tail - 1 ; j >= 0 ; --j )
{
//if ( j < tail - 1 )
// continue ;
chosenCnt = 0 ;
//if ( queue[j].a == 0 )
// printf( "Dummy: %d %d %d\n", j, queue[j].b, 1 - queue[j].b ) ;
SearchDangling( queue[j].a, queue[j].b, used, danglingTime, danglingVisitTime, contigGraph, false, chosen, chosenDummy, chosenCnt, genome ) ;
++danglingTime ;
int prevTag = prevAdd[j] ;
/*if ( queue[j].a == 0 )
{
struct _pair neighbors[5] ;
int ncnt = contigGraph.GetNeighbors( queue[j].a, 1 - queue[j].b, neighbors, 5 ) ;
printf( "%d %d %d %d: %d %d\n", queue[j].b, chosenCnt, prevTag, ncnt, neighbors[0].a, used[ neighbors[0].a ] ) ;
}*/
if ( prevTag == -1 )
break ;
// Trim the dangling list
int k = chosenCnt - 1 ;
if ( j > 0 && j < tail - 1 )
{
for ( k = chosenCnt - 1 ; k >= 1 ; --k )
if ( genome.GetChrIdFromContigId( chosen[k] ) != genome.GetChrIdFromContigId( chosen[k - 1] ) )
break ;
}
// Test the gap size
int len = 0 ;
for ( int l = 0 ; l <= k ; ++l )
{
struct _contig c = genome.GetContigInfo( chosen[k] ) ;
len += c.end - c.start + 1 ;
}
if ( j < tail - 1 )
{
int l ;
for ( l = j ; l >= 0 ; --l )
if ( genome.GetChrIdFromContigId( queue[l].a ) == genome.GetChrIdFromContigId( from ) )
break ;
if ( !ignoreGap && len >= gapSize[ queue[l].a ] + 100 )
continue ;
else
gapSize[ queue[l].a ] -= len ;
}
for ( ; k >= 0 ; --k )
{
used[ chosen[k] ] = true ;
//printf( "Dangling 1: %d=>%d\n", queue[prevTag].a, chosen[k] ) ;
scaffold.InsertNode( queue[ prevTag ].a, 1 - queue[ prevTag ].b, chosen[k], chosenDummy[k] ) ;
}
}
// 3'->5' dangling
for ( int j = 0 ; j < tail ; ++j )
{
//if ( j > 0 )
// continue ;
chosenCnt = 0 ;
SearchDangling( queue[j].a, 1 - queue[j].b, used, danglingTime, danglingVisitTime, contigGraph, false, chosen, chosenDummy, chosenCnt, genome ) ;
++danglingTime ;
int prevTag = prevAdd[j] ;
int nextTag ;
if ( prevTag == -1 || j <= firstAdd )
nextTag = firstAdd ;
else if ( j == prevTag )
nextTag = j ;
else
nextTag = nextAdd[ prevTag ] ;
if ( nextTag == -1 )
break ;
/*if ( queue[j].a == 37549 )
{
struct _pair neighbors[5] ;
int ncnt = contigGraph.GetNeighbors( queue[j].a, queue[j].b, neighbors, 5 ) ;
fprintf( stderr, "%d %d %d: %d %d %d: %d %d %d\n", j, queue[j].a, queue[j].b, chosenCnt, nextTag, ncnt, chosen[0], chosenDummy[0], used[ chosen[0] ] ) ;
}*/
// trim the danling list
int k = chosenCnt - 1 ;
if ( j < tail - 1 && j > 0 )
{
for ( k = chosenCnt - 1 ; k >= 1 ; --k )
if ( genome.GetChrIdFromContigId( chosen[k] ) != genome.GetChrIdFromContigId( chosen[k - 1] ) )
break ;
}
// Test the gap size
int len = 0 ;
for ( int l = 0 ; l <= k ; ++l )
{
struct _contig c = genome.GetContigInfo( chosen[k] ) ;
len += c.end - c.start + 1 ;
}
if ( j > 0 )
{
int l ;
for ( l = j - 1 ; l >= 0 ; --l ) // Notice the j-1 here, because we want the gap strictly before current contig
if ( genome.GetChrIdFromContigId( queue[l].a ) == genome.GetChrIdFromContigId( from ) )
break ;
if ( !ignoreGap && len >= gapSize[ queue[l].a ] + 100 )
continue ;
else
gapSize[ queue[l].a ] -= len ;
}
for ( ; k >= 0 ; --k )
{
used[ chosen[k] ] = true ;
scaffold.InsertNode( queue[nextTag].a, queue[nextTag].b, chosen[k], chosenDummy[k] ) ;
//printf( "Dangling 2: %d<=%d\n", queue[nextTag].a, chosen[k] ) ;
//if ( chosen[k] == 10246 )
// printf( "hi %d %d %d %d\n", j, queue[j].a, k, chosen[k] ) ;
}
}
}
//return 0 ;
// Output the scaffold
int id = 0 ;
char infoFileName[512] ;
char outputFileName[512] ;
sprintf( infoFileName, "%s.info", prefix ) ;
sprintf( outputFileName, "%s.fa", prefix ) ;
outputFile = fopen( outputFileName, "w" ) ;
infoFile = fopen( infoFileName, "w") ;
memset( used, false, sizeof( bool ) * contigCnt ) ;
for ( i = 0 ; i < contigCnt ; ++i )
{
//printf( "%d (%s)\n", i, alignments.GetChromName( genome.GetChrIdFromContigId( i ) ) ) ; fflush( stdout ) ;
/*if ( i == 10246 )
{
std::vector<struct _pair> neighbors ;
scaffold.GetNeighbors( i, 0, neighbors ) ;
printf( "%u\n", neighbors.size() ) ;
}*/
if ( used[i] )
continue ;
int ncnt1 = scaffold.GetNeighbors( i, 0, neighbors, MAX_NEIGHBOR ) ;
int ncnt2 = scaffold.GetNeighbors( i, 1, neighbors, MAX_NEIGHBOR ) ;
if ( ncnt1 == 0 || ncnt2 == 0 ) // The end of a scaffold
{
fprintf( outputFile, ">scaffold_%d\n", id) ;
fprintf( infoFile, ">scaffold_%d", id ) ;
++id ;
int p = i ;
int dummyP = 1 ;
if ( ncnt1 == 0 )
dummyP = 0 ;
used[i] = true ;
genome.PrintContig( outputFile, i, dummyP ) ;
fprintf( infoFile, " (%s %d %c)", alignments.GetChromName( genome.GetChrIdFromContigId( p ) ), p, dummyP == 0 ? '+' : '-' ) ;
while ( 1 )
{
ncnt = scaffold.GetNeighbors( p, 1 - dummyP, neighbors, MAX_NEIGHBOR ) ;
if ( ncnt == 0 )
break ;
// ncnt must be 1
int insertN = 17 ;
if ( genome.GetChrIdFromContigId( p ) == genome.GetChrIdFromContigId( neighbors[0].a ) )
{
struct _contig cp, cna ;
cp = genome.GetContigInfo( p ) ;
cna = genome.GetContigInfo( neighbors[0].a ) ;
if ( p < neighbors[0].a )
insertN = cna.start - cp.end - 1 ;
else if ( p > neighbors[0].a )
insertN = cp.start - cna.end - 1 ;
}
p = neighbors[0].a ;
dummyP = neighbors[0].b ;
for ( int j = 0 ; j < insertN ; ++j )
fprintf( outputFile, "N" ) ;
used[p] = true ;
genome.PrintContig( outputFile, p, dummyP ) ;
fprintf( infoFile, " (%s %d %c)", alignments.GetChromName( genome.GetChrIdFromContigId( p ) ), p, dummyP == 0 ? '+' : '-' ) ;
}
fprintf( outputFile, "\n" ) ;
fprintf( infoFile, "\n" ) ;
}
}
for ( i = 0 ; i < contigCnt ; ++i )
if ( !used[i] )
{
fprintf( stderr, "Unreported contig %d.\n", i ) ;
}
fclose( outputFile ) ;
fclose( infoFile ) ;
delete[] buffer ;
delete[] buffer2 ;
delete[] chosen ;
delete[] queue ;
delete[] counter ;
delete[] visitTime ;
delete[] used ;
delete[] scafInfo ;
delete[] isInQueue ;
delete[] gapSize ;
//fclose( rascafFile ) ;
return 0 ;
}
