static void delete1contig(unsigned int edgeid)
{
edge_array[edgeid].cvg=0;
edge_array[edgeid].deleted=1;
edge_array[edgeid].length=0;
ARC *arc=edge_array[edgeid].arcs;
while(arc)
{
arc->multiplicity=0;
arc->bal_arc->multiplicity=0;
arc=arc->next;
}
if(EdSameAsTwin(edgeid))
return;
edge_array[getTwinEdge(edgeid)].cvg=0;
edge_array[getTwinEdge(edgeid)].deleted=1;
edge_array[getTwinEdge(edgeid)].length=0;
arc = edge_array[getTwinEdge(edgeid)].arcs;
while(arc)
{
arc->multiplicity=0;
arc->bal_arc->multiplicity=0;
arc=arc->next;
}
}
static void moveArc2cp ( unsigned int leftEd, unsigned int rightEd,
unsigned int source, unsigned int target )
{
unsigned int bal_left = getTwinEdge ( leftEd );
unsigned int bal_right = getTwinEdge ( rightEd );
unsigned int bal_source = getTwinEdge ( source );
unsigned int bal_target = getTwinEdge ( target );
ARC * arc;
ARC * newArc, *twinArc;
//between left and source
arc = getArcBetween ( leftEd, source );
arc->to_ed = 0;
newArc = allocateArc ( target );
newArc->multiplicity = arc->multiplicity;
newArc->prev = NULL;
newArc->next = edge_array[leftEd].arcs;
if ( edge_array[leftEd].arcs )
{ edge_array[leftEd].arcs->prev = newArc; }
edge_array[leftEd].arcs = newArc;
arc = getArcBetween ( bal_source, bal_left );
arc->to_ed = 0;
twinArc = allocateArc ( bal_left );
twinArc->multiplicity = arc->multiplicity;
twinArc->prev = NULL;
twinArc->next = NULL;
edge_array[bal_target].arcs = twinArc;
newArc->bal_arc = twinArc;
twinArc->bal_arc = newArc;
//between source and right
arc = getArcBetween ( source, rightEd );
arc->to_ed = 0;
newArc = allocateArc ( rightEd );
newArc->multiplicity = arc->multiplicity;
newArc->prev = NULL;
newArc->next = NULL;
edge_array[target].arcs = newArc;
arc = getArcBetween ( bal_right, bal_source );
arc->to_ed = 0;
twinArc = allocateArc ( bal_target );
twinArc->multiplicity = arc->multiplicity;
twinArc->prev = NULL;
twinArc->next = edge_array[bal_right].arcs;
if ( edge_array[bal_right].arcs )
{ edge_array[bal_right].arcs->prev = twinArc; }
edge_array[bal_right].arcs = twinArc;
newArc->bal_arc = twinArc;
twinArc->bal_arc = newArc;
}
Idx VoronoiDiagram<CoordT>::addVertex(
const PointT& pos, Idx edge1, Idx edge2, Idx edge3
) {
Idx vertex = vertex_pos_.add(pos);
edges_[edge1].end_vertex = vertex;
edges_[edge2].end_vertex = vertex;
edges_[edge3].end_vertex = vertex;
consecutiveEdges(edge1, getTwinEdge(edge3));
consecutiveEdges(edge2, getTwinEdge(edge1));
consecutiveEdges(edge3, getTwinEdge(edge2));
return vertex;
}
void destroyEdge (unsigned int edgeid)
{
unsigned int bal_ed = getTwinEdge (edgeid);
ARC *arc;
if (bal_ed == edgeid)
{
edge_array[edgeid].length = 0;
return;
}
arc = edge_array[edgeid].arcs;
while (arc)
{
arc->bal_arc->to_ed = 0;
arc = arc->next;
}
arc = edge_array[bal_ed].arcs;
while (arc)
{
arc->bal_arc->to_ed = 0;
arc = arc->next;
}
edge_array[edgeid].arcs = NULL;
edge_array[bal_ed].arcs = NULL;
edge_array[edgeid].length = 0;
edge_array[bal_ed].length = 0;
edge_array[edgeid].deleted = 1;
edge_array[bal_ed].deleted = 1;
//printf("Destroyed %d and %d\n",edgeid,bal_ed);
}
static unsigned int deleteLightFlowArc(double min_arc_rate)
{
unsigned int index,twin,count=0;
unsigned int total_in_weight,total_out_weight,coverage;
ARC * arc,*next_arc,*twin_arc;
unsigned int to_ed,twin_te;
for(index=1; index<=num_ed; index++)
{
total_in_weight=0;
total_out_weight=0;
twin = getTwinEdge(index);
coverage = (double)edge_array[index].cvg/10;
arc = edge_array[index].arcs;
while(arc)
{
total_out_weight += arc->multiplicity;
arc=arc->next;
}
arc = edge_array[twin].arcs;
while(arc)
{
total_in_weight += arc->multiplicity;
arc=arc->next;
}
arc = edge_array[index].arcs;
while(arc)
{
next_arc = arc->next;
to_ed = arc ->to_ed;
if(arc->multiplicity != 0 && arc->multiplicity <= (double)total_in_weight*min_arc_rate || arc->multiplicity <= (double)coverage*min_arc_rate)
{
twin_arc=arc->bal_arc;
arc->multiplicity=0;
twin_arc->multiplicity=0;
count++;
}
arc=next_arc;
}
arc = edge_array[twin].arcs;
while(arc)
{
next_arc = arc->next;
to_ed = arc ->to_ed;
if(arc->multiplicity != 0 && arc->multiplicity <= (double)total_out_weight*min_arc_rate || arc->multiplicity <= (double)coverage*min_arc_rate)
{
twin_arc=arc->bal_arc;
arc->multiplicity=0;
twin_arc->multiplicity=0;
count++;
}
arc=next_arc;
}
if(twin != index)
index++;
}
return count;
}
/*************************************************
Function:
compactEdgeArray
Description:
Compacts the edge array by removing deleted edges.
Input:
None.
Output:
None.
Return:
None.
*************************************************/
void compactEdgeArray ()
{
unsigned int i;
unsigned int validCounter = 0;
unsigned int bal_ed;
fprintf ( stderr, "Before compacting, %d edge(s) existed.\n", num_ed );
for ( i = 1; i <= num_ed; i++ )
{
if ( edge_array[i].deleted )
{
continue;
}
validCounter++;
if ( i == validCounter )
{
continue;
}
bal_ed = getTwinEdge ( i );
edgeMove ( i, validCounter );
if ( bal_ed != i )
{
i++;
validCounter++;
}
}
num_ed = validCounter;
fprintf ( stderr, "After compacting, %d edge(s) left.\n", num_ed );
}
void removeLowCovEdges (int lenCutoff, unsigned short covCutoff)
{
unsigned int bal_ed;
unsigned int arcRight_n, arcLeft_n;
ARC *arcLeft, *arcRight;
unsigned int i;
int counter = 0;
for (i = 1; i <= num_ed; i++)
{
if (edge_array[i].deleted || edge_array[i].cvg == 0 || edge_array[i].cvg > covCutoff * 10 || edge_array[i].length >= lenCutoff || EdSameAsTwin (i) || edge_array[i].length == 0)
{
continue;
}
bal_ed = getTwinEdge (i);
arcRight = arcCount (i, &arcRight_n);
arcLeft = arcCount (bal_ed, &arcLeft_n);
if (arcLeft_n < 1 || arcRight_n < 1)
{
continue;
}
destroyEdge (i);
counter++;
}
printf ("Remove low coverage(%d): %d inner edges destroyed\n", covCutoff, counter);
removeDeadArcs ();
linearConcatenate ();
compactEdgeArray ();
}
void compactEdgeArray()
{
unsigned int i;
unsigned int validCounter = 0;
unsigned int bal_ed;
printf ( "there're %d edges\n", num_ed );
for ( i = 1; i <= num_ed; i++ )
{
if ( edge_array[i].deleted )
{ continue; }
validCounter++;
if ( i == validCounter )
{ continue; }
bal_ed = getTwinEdge ( i );
edgeMove ( i, validCounter );
if ( bal_ed != i )
{
i++;
validCounter++;
}
}
num_ed = validCounter;
printf ( "after compacting %d edges left\n", num_ed );
}
static boolean interferingCheck ( unsigned int edgeno, int repTimes )
{
int i, j, t;
unsigned int bal_ed;
involved[0] = edgeno;
i = 1;
for ( j = 0; j < repTimes; j++ )
{ involved[i++] = lefts[j]; }
for ( j = 0; j < repTimes; j++ )
{ involved[i++] = rights[j]; }
for ( j = 0; j < i - 1; j++ )
for ( t = j + 1; t < i; t++ )
if ( involved[j] == involved[t] )
{ return 1; }
for ( j = 0; j < i; j++ )
{
bal_ed = getTwinEdge ( involved[j] );
for ( t = 0; t < i; t++ )
if ( bal_ed == involved[t] )
{ return 1; }
}
return 0;
}
void output_graph ( char * outfile )
{
char name[256];
FILE * fp;
unsigned int i, bal_i;
sprintf ( name, "%s.edge.gvz", outfile );
fp = ckopen ( name, "w" );
fprintf ( fp, "digraph G{\n" );
fprintf ( fp, "\tsize=\"512,512\";\n" );
for ( i = num_ed; i > 0; i-- )
{
if ( edge_array[i].deleted )
{ continue; }
/*
arcCount(i,&arcNum);
if(arcNum<1)
continue;
*/
bal_i = getTwinEdge ( i );
/*
arcCount(bal_i,&arcNum);
if(arcNum<1)
continue;
*/
fprintf ( fp, "\tV%d -> V%d[label =\"%d(%d)\"];\n", edge_array[i].from_vt, edge_array[i].to_vt, i, edge_array[i].length );
}
fprintf ( fp, "}\n" );
fclose ( fp );
}
//a path from e1 to e2 is merged int to e1(indicate=0) or e2(indicate=1), update graph topology
void linearUpdateConnection (unsigned int e1, unsigned int e2, int indicate)
{
unsigned int bal_ed;
ARC *parc;
if (!indicate)
{
edge_array[e1].to_vt = edge_array[e2].to_vt;
bal_ed = getTwinEdge (e1);
parc = edge_array[e2].arcs;
while (parc)
{
parc->bal_arc->to_ed = bal_ed;
parc = parc->next;
}
edge_array[e1].arcs = edge_array[e2].arcs;
edge_array[e2].arcs = NULL;
if (edge_array[e1].length || edge_array[e2].length)
edge_array[e1].cvg = (edge_array[e1].cvg * edge_array[e1].length + edge_array[e2].cvg * edge_array[e2].length) / (edge_array[e1].length + edge_array[e2].length);
edge_array[e2].deleted = 1;
}
else
{
//all the arcs pointing to e1 switch to e2
parc = edge_array[getTwinEdge (e1)].arcs;
while (parc)
{
parc->bal_arc->to_ed = e2;
parc = parc->next;
}
edge_array[e1].arcs = NULL;
edge_array[e2].from_vt = edge_array[e1].from_vt;
if (edge_array[e1].length || edge_array[e2].length)
edge_array[e2].cvg = (edge_array[e1].cvg * edge_array[e1].length + edge_array[e2].cvg * edge_array[e2].length) / (edge_array[e1].length + edge_array[e2].length);
edge_array[e1].deleted = 1;
}
}
/*************************************************
Function:
add1marker2edge
Description:
Records the id of read which crosses the edge.
Input:
1. edgeno: the edge index
2. readid: the read id
Output:
None.
Return:
None.
*************************************************/
static void add1marker2edge ( unsigned int edgeno, long long readid )
{
if ( edge_array[edgeno].multi == 255 )
{
return;
}
unsigned int bal_ed = getTwinEdge ( edgeno );
unsigned char counter = edge_array[edgeno].multi++;
edge_array[edgeno].markers[counter] = readid;
counter = edge_array[bal_ed].multi++;
edge_array[bal_ed].markers[counter] = -readid;
}
static unsigned int cp1edge ( unsigned int source, unsigned int target )
{
int length = edge_array[source].length;
char * tightSeq;
int index;
unsigned int bal_source = getTwinEdge ( source );
unsigned int bal_target;
if ( bal_source > source )
{ bal_target = target + 1; }
else
{
bal_target = target;
target = target + 1;
}
tightSeq = ( char * ) ckalloc ( ( length / 4 + 1 ) * sizeof ( char ) );
for ( index = 0; index < length / 4 + 1; index++ )
{ tightSeq[index] = edge_array[source].seq[index]; }
edge_array[target].length = length;
edge_array[target].cvg = edge_array[source].cvg;
edge_array[target].to_vt = edge_array[source].to_vt;
edge_array[target].from_vt = edge_array[source].from_vt;
edge_array[target].seq = tightSeq;
edge_array[target].bal_edge = edge_array[source].bal_edge;
edge_array[target].rv = NULL;
edge_array[target].arcs = NULL;
edge_array[target].markers = NULL;
edge_array[target].flag = 0;
edge_array[target].deleted = 0;
tightSeq = ( char * ) ckalloc ( ( length / 4 + 1 ) * sizeof ( char ) );
for ( index = 0; index < length / 4 + 1; index++ )
{ tightSeq[index] = edge_array[bal_source].seq[index]; }
edge_array[bal_target].length = length;
edge_array[bal_target].cvg = edge_array[bal_source].cvg;
edge_array[bal_target].to_vt = edge_array[bal_source].to_vt;
edge_array[bal_target].from_vt = edge_array[bal_source].from_vt;
edge_array[bal_target].seq = tightSeq;
edge_array[bal_target].bal_edge = edge_array[bal_source].bal_edge;
edge_array[bal_target].rv = NULL;
edge_array[bal_target].arcs = NULL;
edge_array[bal_target].markers = NULL;
edge_array[bal_target].flag = 0;
edge_array[bal_target].deleted = 0;
return target;
}
//move edge from source to target
void edgeMove (unsigned int source, unsigned int target)
{
unsigned int bal_source, bal_target;
ARC *arc;
copyEdge (source, target);
bal_source = getTwinEdge (source);
//bal_edge
if (bal_source != source)
{
bal_target = target + 1;
copyEdge (bal_source, bal_target);
edge_array[target].bal_edge = 2;
edge_array[bal_target].bal_edge = 0;
}
else
{
edge_array[target].bal_edge = 1;
bal_target = target;
}
//take care of the arcs
arc = edge_array[target].arcs;
while (arc)
{
arc->bal_arc->to_ed = bal_target;
arc = arc->next;
}
if (bal_target == target)
{
return;
}
arc = edge_array[bal_target].arcs;
while (arc)
{
arc->bal_arc->to_ed = target;
arc = arc->next;
}
}
void delowHighArc(int multi)
{
unsigned int i, twin,to_edge,count = 0;
ARC *arc, *arc_temp;
unsigned int in_weight,out_weight,curr_weight;
for (i = 1; i <= num_ed; i++)
{
in_weight=0;
curr_weight=0;
//获取i的in_flow权重
twin=getTwinEdge(i);
arc=edge_array[twin].arcs;
while(arc)
{
in_weight += arc->multiplicity;
arc=arc->next;
}
arc = edge_array[i].arcs;
while (arc)
{
curr_weight = arc->multiplicity;
to_edge = arc->to_ed;
arc_temp = edge_array[to_edge].arcs;
out_weight=0;
while(arc_temp)
{
out_weight += arc_temp->multiplicity;
arc_temp=arc_temp->next;
}
if( in_weight != 0 && curr_weight !=0 && curr_weight > in_weight*multi && curr_weight > out_weight*multi)
{
count++;
arc->multiplicity= in_weight > out_weight ? in_weight : out_weight;
}
arc=arc->next;
}
}
// printf("delow arc : %d\n",count);
}
void removeWeakEdges (int lenCutoff, unsigned int multiCutoff)
{
unsigned int bal_ed;
unsigned int arcRight_n, arcLeft_n;
ARC *arcLeft, *arcRight;
unsigned int i;
int counter = 0;
for (i = 1; i <= num_ed; i++)
{
if (edge_array[i].deleted || edge_array[i].length == 0 || edge_array[i].length > lenCutoff || EdSameAsTwin (i))
{
continue;
}
bal_ed = getTwinEdge (i);
arcRight = arcCount (i, &arcRight_n);
if (arcRight_n > 1 || !arcRight || arcRight->multiplicity > multiCutoff)
{
continue;
}
arcLeft = arcCount (bal_ed, &arcLeft_n);
if (arcLeft_n > 1 || !arcLeft || arcLeft->multiplicity > multiCutoff)
{
continue;
}
destroyEdge (i);
counter++;
}
printf ("%d weak inner edges destroyed\n", counter);
removeDeadArcs ();
/*
linearConcatenate();
compactEdgeArray();
*/
}
static int extern_contig(unsigned int edgeid,int pool_index)
{
if(pool[edgeid]!=0)
return 0;
pool[edgeid]=pool_index;
pool[ getTwinEdge(edgeid)]=pool_index;
int length=0;
length += edge_array[edgeid].length;
ARC *arc;
unsigned int best_id;
int max_arc;
unsigned int curr_edge= edgeid;
while(curr_edge)
{
max_arc=0;
arc = edge_array[edgeid].arcs;
while(arc)
{
if(pool[arc->to_ed] ==0)
{
if(arc->multiplicity > max_arc)
{
max_arc=arc->multiplicity;
best_id=arc->to_ed;
}
}
arc=arc->next;
}
if(max_arc>0)
{
pool[best_id]=pool_index;
pool[getTwinEdge(best_id)]=pool_index;
length += edge_array[best_id].length;
curr_edge=best_id;
}
else
curr_edge=0;
}
curr_edge= getTwinEdge(edgeid);
while(curr_edge)
{
max_arc=0;
arc = edge_array[edgeid].arcs;
while(arc)
{
if(pool[arc->to_ed] ==0)
{
if(arc->multiplicity > max_arc)
{
max_arc=arc->multiplicity;
best_id=arc->to_ed;
}
}
arc=arc->next;
}
if(max_arc>0)
{
pool[best_id]=pool_index;
pool[getTwinEdge(best_id)]=pool_index;
length += edge_array[best_id].length;
curr_edge=best_id;
}
else
curr_edge=0;
}
return length;
}
void output_contig ( EDGE * ed_array, unsigned int ed_num, char * outfile, int cut_len )
{
char temp[256];
FILE * fp, *fp_contig;
int flag, count, len_c;
int signI;
unsigned int i;
long long sum = 0, N90, N50;
unsigned int * length_array;
boolean tip;
sprintf ( temp, "%s.contig", outfile );
fp = ckopen ( temp, "w" );
qsort ( &ed_array[1], ed_num, sizeof ( EDGE ), cmp_edge );
length_array = ( unsigned int * ) ckalloc ( ed_num * sizeof ( unsigned int ) );
kmerSeq = ( char * ) ckalloc ( overlaplen * sizeof ( char ) );
//first scan for number counting
count = len_c = 0;
for ( i = 1; i <= ed_num; i++ )
{
if ( ( ed_array[i].length + overlaplen ) >= len_bar )
{ length_array[len_c++] = ed_array[i].length + overlaplen; }
if ( ed_array[i].length < 1 || ed_array[i].deleted )
{ continue; }
count++;
if ( EdSmallerThanTwin ( i ) )
{ i++; }
}
sum = 0;
for ( signI = len_c - 1; signI >= 0; signI-- )
{ sum += length_array[signI]; }
if ( len_c > 0 )
{ printf ( "%d ctgs longer than %d, sum up %lldbp, with average length %lld\n", len_c, len_bar, sum, sum / len_c ); }
qsort ( length_array, len_c, sizeof ( length_array[0] ), cmp_int );
printf ( "the longest is %dbp, ", length_array[len_c - 1] );
N50 = sum * 0.5;
N90 = sum * 0.9;
sum = flag = 0;
for ( signI = len_c - 1; signI >= 0; signI-- )
{
sum += length_array[signI];
if ( !flag && sum >= N50 )
{
printf ( "contig N50 is %d bp,", length_array[signI] );
flag = 1;
}
if ( sum >= N90 )
{
printf ( "contig N90 is %d bp\n", length_array[signI] );
break;
}
}
//fprintf(fp,"Number %d\n",count);
for ( i = 1; i <= ed_num; i++ )
{
//if(ed_array[i].multi!=1||ed_array[i].length<1||(ed_array[i].length+overlaplen)<cut_len)
if ( ed_array[i].deleted || ed_array[i].length < 1 )
{ continue; }
if ( ed_array[i].arcs && ed_array[getTwinEdge ( i )].arcs )
{ tip = 0; }
else
{ tip = 1; }
output_1contig ( i, & ( ed_array[i] ), fp, tip );
if ( EdSmallerThanTwin ( i ) )
{ i++; }
}
fclose ( fp );
free ( ( void * ) kmerSeq );
free ( ( void * ) length_array );
printf ( "%d contigs longer than %d output\n", count, cut_len );
sprintf ( temp, "%s.ContigIndex", outfile );
fp_contig = ckopen ( temp, "w" );
fprintf ( fp_contig, "Edge_num %d %d\n", ed_num, count );
fprintf ( fp_contig, "index\tlength\treverseComplement\n" );
for ( i = 1; i <= num_ed; i++ )
{
fprintf ( fp_contig, "%d\t%d\t", i, edge_array[i].length + overlaplen );
if ( EdSmallerThanTwin ( i ) )
{
fprintf ( fp_contig, "1\n" );
i++;
}
else if ( EdLargerThanTwin ( i ) )
{ fprintf ( fp_contig, "-1\n" ); }
else
{ fprintf ( fp_contig, "0\n" ); }
}
fclose ( fp_contig );
}
/*************************************************
Function:
loadPath
Description:
1. Loads the path info.
2. Records the ids of reads crossing edges.
Input:
1. graphfile: the input prefix
Output:
None.
Return:
None.
*************************************************/
boolean loadPath ( char *graphfile )
{
FILE *fp;
char name[256], line[1024];
unsigned int i, bal_ed, num1, edgeno, num2;
long long markCounter = 0, readid = 0;
char *seg;
sprintf ( name, "%s.markOnEdge", graphfile );
fp = fopen ( name, "r" );
if ( !fp )
{
return 0;
}
for ( i = 1; i <= num_ed; i++ )
{
edge_array[i].multi = 0;
}
for ( i = 1; i <= num_ed; i++ )
{
fscanf ( fp, "%d", &num1 );
if ( EdSmallerThanTwin ( i ) )
{
fscanf ( fp, "%d", &num2 );
bal_ed = getTwinEdge ( i );
if ( num1 + num2 >= 255 )
{
edge_array[i].multi = 255;
edge_array[bal_ed].multi = 255;
}
else
{
edge_array[i].multi = num1 + num2;
edge_array[bal_ed].multi = num1 + num2;
markCounter += 2 * ( num1 + num2 );
}
i++;
}
else
{
if ( 2 * num1 >= 255 )
{
edge_array[i].multi = 255;
}
else
{
edge_array[i].multi = 2 * num1;
markCounter += 2 * num1;
}
}
}
fclose ( fp );
fprintf ( stderr, "%lld markers overall.\n", markCounter );
markersArray = ( long long * ) ckalloc ( markCounter * sizeof ( long long ) );
markCounter = 0;
for ( i = 1; i <= num_ed; i++ )
{
if ( edge_array[i].multi == 255 )
{
continue;
}
edge_array[i].markers = markersArray + markCounter;
markCounter += edge_array[i].multi;
edge_array[i].multi = 0;
}
sprintf ( name, "%s.path", graphfile );
fp = fopen ( name, "r" );
if ( !fp )
{
return 0;
}
while ( fgets ( line, sizeof ( line ), fp ) != NULL )
{
//printf("%s",line);
readid++;
seg = strtok ( line, " " );
while ( seg )
{
edgeno = atoi ( seg );
//printf("%s, %d\n",seg,edgeno);
add1marker2edge ( edgeno, readid );
seg = strtok ( NULL, " " );
}
}
fclose ( fp );
markCounter = 0;
for ( i = 1; i <= num_ed; i++ )
{
if ( edge_array[i].multi == 255 )
{
continue;
}
markCounter += edge_array[i].multi;
}
fprintf ( stderr, "%lld marks loaded.\n", markCounter );
return 1;
}
/*
- -
> - <
- -
*/
unsigned int solvable ( unsigned int edgeno )
{
if ( EdSameAsTwin ( edgeno ) || edge_array[edgeno].multi == 255 )
{ return 0; }
unsigned int bal_ed = getTwinEdge ( edgeno );
unsigned int arcRight_n, arcLeft_n;
unsigned int counter;
unsigned int i, j;
unsigned int branch, bal_branch;
ARC * parcL, *parcR;
parcL = arcCounts ( bal_ed, &arcLeft_n );
if ( arcLeft_n < 2 )
{ return 0; }
parcR = arcCounts ( edgeno, &arcRight_n );
if ( arcLeft_n != arcRight_n )
{ return 0; }
// check each right branch only has one upsteam connection
/*
if(edgeno==2551){
for(i=0;i<arcLeft_n;i++)
printf("%d,",lefts[i]);
printf("__left to %d\n",edgeno);
for(j=0;j<arcRight_n;j++)
printf("%d,",rights[j]);
printf("__right to %d\n",edgeno);
}
*/
arcRight_n = 0;
while ( parcR )
{
if ( parcR->to_ed == 0 )
{
parcR = parcR->next;
continue;
}
branch = parcR->to_ed;
if ( EdSameAsTwin ( branch ) || edge_array[branch].multi == 255 )
{
return 0;
}
rights[arcRight_n++] = branch;
bal_branch = getTwinEdge ( branch );
arcCounts ( bal_branch, &counter );
if ( counter != 1 )
{
return 0;
}
parcR = parcR->next;
}
// check if each left branch only has one downsteam connection
arcLeft_n = 0;
while ( parcL )
{
if ( parcL->to_ed == 0 )
{
parcL = parcL->next;
continue;
}
branch = parcL->to_ed;
if ( EdSameAsTwin ( branch ) || edge_array[branch].multi == 255 )
{ return 0; }
bal_branch = getTwinEdge ( branch );
lefts[arcLeft_n++] = bal_branch;
arcCounts ( bal_branch, &counter );
if ( counter != 1 )
{ return 0; }
parcL = parcL->next;
}
//check if reads indicate one to one connection between upsteam and downstream edges
for ( i = 0; i < arcLeft_n; i++ )
{
counter = 0;
for ( j = 0; j < arcRight_n; j++ )
{
gothrough[i][j] = cntByReads ( lefts[i], edgeno, rights[j] ) == 0 ? 0 : 1;
counter += gothrough[i][j];
if ( counter > 1 )
{ return 0; }
}
if ( counter != 1 )
{ return 0; }
}
for ( j = 0; j < arcRight_n; j++ )
{
counter = 0;
for ( i = 0; i < arcLeft_n; i++ )
{ counter += gothrough[i][j]; }
if ( counter != 1 )
{ return 0; }
}
return arcLeft_n;
}
boolean isUnreliableTip_strict (unsigned int edgeid, int cutLen)
{
unsigned int arcRight_n, arcLeft_n;
unsigned int bal_ed;
unsigned int currentEd = edgeid;
int length = 0;
unsigned int mult = 0;
ARC *arc, *activeArc = NULL, *tempArc;
if (edgeid == 0)
{
return 0;
}
bal_ed = getTwinEdge (edgeid);
if (bal_ed == edgeid)
{
return 0;
}
arcCount (bal_ed, &arcLeft_n);
if (arcLeft_n > 0)
{
return 0;
}
while (currentEd)
{
arcCount (bal_ed, &arcLeft_n);
tempArc = arcCount (currentEd, &arcRight_n);
if (arcLeft_n > 1 || arcRight_n > 1)
{
if (arcLeft_n == 0 || length == 0)
{
return 0;
}
else
{
break;
}
}
length += edge_array[currentEd].length;
if (length >= cutLen)
{
return 0;
}
if (tempArc)
{
activeArc = tempArc;
currentEd = activeArc->to_ed;
bal_ed = getTwinEdge (currentEd);
}
else
{
currentEd = 0;
}
}
if (currentEd == 0)
{
caseA++;
return 1;
}
if (!activeArc)
{
printf ("no activeArc while checking edge %d\n", edgeid);
}
if (activeArc->multiplicity == 1)
{
caseB++;
return 1;
}
for (arc = edge_array[bal_ed].arcs; arc != NULL; arc = arc->next)
if (arc->multiplicity > mult)
{
mult = arc->multiplicity;
}
if (mult > activeArc->multiplicity)
{
caseC++;
}
return mult > activeArc->multiplicity;
}
boolean loadPathBin ( char *graphfile )
{
FILE *fp;
char name[256];
unsigned int i, bal_ed, num1, num2;
long long markCounter = 0, readid = 0;
unsigned char seg, ch;
unsigned int *freadBuf;
sprintf ( name, "%s.markOnEdge", graphfile );
fp = fopen ( name, "r" );
if ( !fp )
{
return 0;
}
for ( i = 1; i <= num_ed; i++ )
{
edge_array[i].multi = 0;
edge_array[i].markers = NULL;
}
for ( i = 1; i <= num_ed; i++ )
{
fscanf ( fp, "%d", &num1 );
if ( EdSmallerThanTwin ( i ) )
{
fscanf ( fp, "%d", &num2 );
bal_ed = getTwinEdge ( i );
if ( num1 + num2 >= 255 )
{
edge_array[i].multi = 255;
edge_array[bal_ed].multi = 255;
}
else
{
edge_array[i].multi = num1 + num2;
edge_array[bal_ed].multi = num1 + num2;
markCounter += 2 * ( num1 + num2 );
}
i++;
}
else
{
if ( 2 * num1 >= 255 )
{
edge_array[i].multi = 255;
}
else
{
edge_array[i].multi = 2 * num1;
markCounter += 2 * num1;
}
}
}
fclose ( fp );
fprintf ( stderr, "%lld markers overall.\n", markCounter );
markersArray = ( long long * ) ckalloc ( markCounter * sizeof ( long long ) );
markCounter = 0;
for ( i = 1; i <= num_ed; i++ )
{
if ( edge_array[i].multi == 255 )
{
continue;
}
edge_array[i].markers = markersArray + markCounter;
markCounter += edge_array[i].multi;
edge_array[i].multi = 0;
}
sprintf ( name, "%s.path", graphfile );
fp = fopen ( name, "rb" );
if ( !fp )
{
return 0;
}
freadBuf = ( unsigned int * ) ckalloc ( ( maxReadLen - overlaplen + 1 ) * sizeof ( unsigned int ) );
while ( fread ( &ch, sizeof ( char ), 1, fp ) == 1 )
{
//printf("%s",line);
if ( fread ( freadBuf, sizeof ( unsigned int ), ch, fp ) != ch )
{
break;
}
readid++;
for ( seg = 0; seg < ch; seg++ )
{
add1marker2edge ( freadBuf[seg], readid );
}
}
fclose ( fp );
markCounter = 0;
for ( i = 1; i <= num_ed; i++ )
{
if ( edge_array[i].multi == 255 )
{
continue;
}
markCounter += edge_array[i].multi;
}
for ( i = 0; i <= num_ed; i++ )
{
if ( edge_array[i].multi >= 2 && edge_array[i].multi != 255 )
{
qsort ( edge_array[i].markers, ( int ) edge_array[i].multi, sizeof ( long long ), comp );
}
}
fprintf ( stderr, "%lld markers loaded.\n", markCounter );
free ( ( void * ) freadBuf );
return 1;
}
//concatenate two edges if they are linearly linked
void linearConcatenate ()
{
unsigned int i;
int conc_c = 1;
int counter;
unsigned int from_ed, to_ed, bal_ed;
ARC *parc, *parc2;
unsigned int bal_fe;
//debugging(30514);
while (conc_c)
{
conc_c = 0;
counter = 0;
for (i = 1; i <= num_ed; i++) //num_ed
{
if (edge_array[i].deleted || EdSameAsTwin (i))
{
continue;
}
if (edge_array[i].length > 0)
{
counter++;
}
parc = edge_array[i].arcs;
if (!parc || parc->next)
{
continue;
}
to_ed = parc->to_ed;
bal_ed = getTwinEdge (to_ed);
parc2 = edge_array[bal_ed].arcs;
if (bal_ed == to_ed || !parc2 || parc2->next)
{
continue;
}
from_ed = i;
if (from_ed == to_ed || from_ed == bal_ed)
{
continue;
}
//linear connection found
conc_c++;
linearUpdateConnection (from_ed, to_ed, 0);
allpathUpdateEdge (from_ed, to_ed, 0);
bal_fe = getTwinEdge (from_ed);
linearUpdateConnection (bal_ed, bal_fe, 1);
allpathUpdateEdge (bal_ed, bal_fe, 1);
/*
if(from_ed==6589||to_ed==6589)
printf("%d <- %d (%d)\n",from_ed,to_ed,i);
if(bal_fe==6589||bal_ed==6589)
printf("%d <- %d (%d)\n",bal_fe,bal_ed,i);
*/
}
printf ("a linear concatenation lap, %d concatenated\n", conc_c);
}
printf ("%d edges in graph\n", counter);
}
int deleteLightContig()
{
double prev_cov,next_cov,max,min,curr_cov;
unsigned int index;
int change=0;
ARC * arc,*arc_temp;
for(index=1; index<=num_ed; index++)
{
if(EdSameAsTwin(index))
continue;
computeNextCov(index,&next_cov);
computeNextCov(getTwinEdge(index),&prev_cov);
if(next_cov ==0 || prev_cov ==0)
continue;
if(next_cov > prev_cov)
{
max=next_cov;
min=prev_cov;
}
else
{
max=prev_cov;
min=next_cov;
}
curr_cov = (double)edge_array[index].cvg;
printf("contig_cov:\t%0.1f\t%0.1f\t%0.1f\n",curr_cov,max,min);
if(min / max <0.1)
{
if(curr_cov /min < 0.5)
{
delete1contig(index);
}
}
else
{
if(curr_cov / max <0.05)
{
delete1contig(index);
}
}
index++;
}
/*
for(index=1;index<=num_ed;index++)
{
arc=edge_array[index].arcs;
while(arc)
{
arc_temp=arc;
arc=arc->next;
if(arc_temp->multiplicity == 0)
{
if(arc_temp->prev)
{
arc_temp->prev->next=arc_temp->next;
}
else
{
edge_array[index].arcs=arc_temp->next;
}
if(arc_temp->next)
{
arc_temp->next->prev = arc_temp->prev;
}
dismissArc (arc_temp);
change++;
}
}
}*/
change = removeArc();
return change>0?1:0;;
}
Idx VoronoiDiagram<CoordT>::getStartVertex(Idx edge) const {
return edges_[getTwinEdge(edge)].end_vertex;
}
void output_contig (EDGE * ed_array, unsigned int ed_num, char *outfile, int cut_len)
{
char temp[256];
FILE * fp, *fp_contig;
int flag, count, len_c;
int signI;
unsigned int i, j, diff_len=0;
long long sum = 0, N90, N50;
unsigned int *length_array;
boolean tip;
sprintf (temp, "%s.contig", outfile);
fp = ckopen (temp, "w");
index_array = (unsigned int *)ckalloc((ed_num+1)*sizeof(unsigned int));
unsigned int * all_length_arr = (unsigned int*) ckalloc((ed_num+1)*sizeof(unsigned int));
flag_array = (unsigned int*)ckalloc((ed_num+1)*sizeof(unsigned int));
for (i=1; i<=ed_num; ++i)
{
index_array[i] = ed_array[i].length;
all_length_arr[i] = ed_array[i].length;
}
qsort(&all_length_arr[1], ed_num, sizeof(all_length_arr[0]), cmp_int);
for (i=1; i<=ed_num; ++i)
{
for (j=i+1; j<=ed_num; ++j)
{
if (all_length_arr[i] != all_length_arr[j])
break;
}
all_length_arr[++diff_len] = all_length_arr[i];
flag_array[diff_len] = i;
i = j-1;
}
for (i=1; i<=ed_num; ++i)
{
index_array[i] = uniqueLenSearch(all_length_arr, flag_array, diff_len, index_array[i]);
}
for (i=1; i<=ed_num; ++i)
{
flag_array[index_array[i]] = i;
}
free((void*)all_length_arr);
length_array = (unsigned int *) ckalloc (ed_num * sizeof (unsigned int));
kmerSeq = (char *) ckalloc (overlaplen * sizeof (char));
count = len_c = 0;
for (i = 1; i <= ed_num; i++)
{
if ((ed_array[i].length + overlaplen) >= len_bar)
{
length_array[len_c++] = ed_array[i].length + overlaplen;
}
if (ed_array[i].length < 1 || ed_array[i].deleted)
{
continue;
}
count++;
if (EdSmallerThanTwin (i))
{
i++;
}
}
sum = 0;
for (signI = len_c - 1; signI >= 0; signI--)
{
sum += length_array[signI];
}
qsort ( length_array, len_c, sizeof ( length_array[0] ), cmp_int );
if ( len_c > 0 )
{
printf ( "%d ctgs longer than %d, sum up %lldbp, with average length %lld\n", len_c, len_bar, sum, sum / len_c );
printf ( "the longest is %dbp, ", length_array[len_c - 1] );
}
N50 = sum * 0.5;
N90 = sum * 0.9;
sum = flag = 0;
for (signI = len_c - 1; signI >= 0; signI--)
{
sum += length_array[signI];
if (!flag && sum >= N50)
{
printf ("contig N50 is %d bp,", length_array[signI]);
flag = 1;
}
if (sum >= N90)
{
printf ("contig N90 is %d bp\n", length_array[signI]);
break;
}
}
for (i = 1; i <= ed_num; i++)
{
j = flag_array[i];
if (ed_array[j].deleted || ed_array[j].length < 1)
{
continue;
}
if (ed_array[j].arcs && ed_array[getTwinEdge (j)].arcs)
{
tip = 0;
}
else
{
tip = 1;
}
output_1contig (i, &(ed_array[j]), fp, tip);
if (EdSmallerThanTwin (j))
{
i++;
}
}
fclose (fp);
free ((void *) kmerSeq);
free ((void *) length_array);
printf ("%d contigs longer than %d output\n", count, cut_len);
sprintf (temp, "%s.ContigIndex", outfile);
fp_contig = ckopen (temp, "w");
fprintf (fp_contig, "Edge_num %d %d\n", ed_num, count);
fprintf (fp_contig, "index\tlength\treverseComplement\n");
for (i = 1; i <= num_ed; i++)
{
j = flag_array[i];
fprintf (fp_contig, "%d\t%d\t", i, edge_array[j].length + overlaplen);
if (EdSmallerThanTwin (j))
{
fprintf (fp_contig, "1\n");
i++;
}
else if (EdLargerThanTwin (j))
{
fprintf (fp_contig, "-1\n");
}
else
{
fprintf (fp_contig, "0\n");
}
}
fclose (fp_contig);
}