void estimate_distinct_kmers(unsigned long genome_size, Bank *Reads)
{
int size_linearCounter = genome_size * 8; // alloc 8 bits * genome size for counting, i.e. ~ as much as the assembly Bloom size
LinearCounter *linearCounter = new LinearCounter(size_linearCounter);
bloom_pass_reads(Reads,linearCounter, (BloomCpt * ) NULL, (char*)"%cEstimating number of distinct kmers (%lld reads processed so far)");
long nb_distinct_kmers = linearCounter->count();
if (linearCounter->is_accurate())
printf("Estimated that %ld distinct kmers are in the reads\n",nb_distinct_kmers);
else
printf("Cannot estimate the number of distinct kmers. Allocate a larger counter\n");
delete linearCounter;
}
//------------- Start of function VgaBuf::indicator --------------//
//
// <int> x1, y1, x2, y2 = coordination of the indicator
// <float> curValue = the value of the bar
// <float> maxValue = max value, the bar width = maxBarWidth * curValue / maxValue
// <int> indiColor = color of the indicator
// [int] backColor = background color
// (default : vga.color_down)
// (-2 if don't paint background color)
//
void VgaBuf::indicator(int x1, int y1, int x2, int y2, float curValue,
float maxValue, int indiColor, int backColor)
{
if( backColor == -1 )
backColor = color_down;
int cutPoint;
if( curValue <= 0.0f )
cutPoint = x1;
else if( curValue >= maxValue )
cutPoint = x2 + 1;
else
cutPoint = x1 + int(float(x2 - x1 + 1) * curValue / maxValue);
if( cutPoint > x1 )
{
int cutHeight = (y1 * 5 + y2 * 3) / 8; // cut at 3/8 of between y1 and y2
err_when( y2 - y1 - 1 < 4 );
LinearCounter brightness;
brightness.init( y1, 0, cutHeight, MAX_BRIGHTNESS_ADJUST_DEGREE-2 ); // glowing
int y;
for( y=y1; y <= cutHeight; y++, brightness.inc() )
{
barW_fast( x1, y, cutPoint, y, vga.vga_color_table->get_table(brightness.y)[indiColor]);
}
brightness.init( cutHeight, MAX_BRIGHTNESS_ADJUST_DEGREE-2, y2, -MAX_BRIGHTNESS_ADJUST_DEGREE/2 ); //
for( ; y<=y2; ++y, brightness.inc() )
{
barW_fast( x1, y, cutPoint, y, vga.vga_color_table->get_table(brightness.y)[indiColor]);
}
}
if( cutPoint <= x2 )
{
bar( cutPoint, y1, x2, y2, backColor );
}
}
//------------- Start of function VgaBuf::indicator --------------//
//
// <int> x1, y1, x2, y2 = coordination of the indicator
// <float> curValue = the value of the bar
// <float> maxValue = max value, the bar width = maxBarWidth * curValue / maxValue
// <int> indiColor = color of the indicator
// [int] backColor = background color
// (default : vga.color_down)
// (-2 if don't paint background color)
//
void VgaBuf::indicator(int x1, int y1, int x2, int y2, float curValue,
float maxValue, int indiColor, int backColor)
{
if( backColor == -1 )
backColor = color_down;
/*
if( curValue > maxValue )
curValue = maxValue;
if( curValue > 0 )
{
int barWidth = (int) ((float)(x2-x1) * curValue / maxValue);
int halfHeight = (y2-y1+1)/2-1;
int tx2 = x1+barWidth;
int y;
indiColor+=halfHeight;
for( y=y1 ; y<y1+halfHeight ; y++, indiColor-- )
bar( x1, y, tx2, y, indiColor );
for( ; y<=y2 ; y++, indiColor++ )
bar( x1, y, tx2, y, indiColor );
if( backColor != -2 ) // -2 if don't paint background color
{
if( x1+barWidth < x2 )
bar( x1+barWidth+1, y1, x2, y2, backColor );
}
}
else
{
if( backColor != -2 ) // -2 if don't paint background color
bar( x1, y1, x2, y2, backColor );
}
*/
int cutPoint;
if( curValue <= 0.0f )
cutPoint = x1;
else if( curValue >= maxValue )
cutPoint = x2 + 1;
else
cutPoint = x1 + int(float(x2 - x1 + 1) * curValue / maxValue);
if( cutPoint > x1 )
{
if( is_front )
mouse.hide_area( x1, y1, cutPoint, y2 );
int cutHeight = (y1 * 5 + y2 * 3) / 8; // cut at 3/8 of between y1 and y2
err_when( y2 - y1 - 1 < 4 );
LinearCounter brightness;
brightness.init( y1, 0, cutHeight, MAX_BRIGHTNESS_ADJUST_DEGREE-2 ); // glowing
int y;
for( y=y1; y <= cutHeight; y++, brightness.inc() )
{
barW_fast( x1, y, cutPoint, y, vga.vga_color_table->get_table(brightness.y)[indiColor]);
}
brightness.init( cutHeight, MAX_BRIGHTNESS_ADJUST_DEGREE-2, y2, -MAX_BRIGHTNESS_ADJUST_DEGREE/2 ); //
for( ; y<=y2; ++y, brightness.inc() )
{
barW_fast( x1, y, cutPoint, y, vga.vga_color_table->get_table(brightness.y)[indiColor]);
}
if( is_front )
mouse.show_area();
}
if( cutPoint <= x2 )
{
bar( cutPoint, y1, x2, y2, backColor );
}
}
uint64_t extrapolate_distinct_kmers_wrapped(unsigned long nbytes_memory, Bank *Reads)
{
unsigned long size_linearCounter = nbytes_memory * 8L; // alloc 8 bits * nbytes for counting
LinearCounter *linearCounter = new LinearCounter(size_linearCounter);
int stops = 100000;
// variant of bloom_pass_reads
int64_t NbRead = 0;
int64_t NbInsertedKmers = 0;
Reads->rewind_all();
char * rseq;
long i;
kmer_type kmer, graine, graine_revcomp;
long nb_distinct_kmers = 0;
long previous_nb_distinct_kmers = 0;
uint64_t estimated_nb_reads = Reads->estimate_nb_reads();
bool stop = false;
while (Reads->get_next_seq(&rseq,&readlen))
{
if (stop)
break;
for (i=0; i<readlen-sizeKmer+1; i++)
{
kmer = extractKmerFromRead(rseq,i,&graine,&graine_revcomp);
linearCounter->add(kmer);
NbInsertedKmers++;
if (NbInsertedKmers % stops == 0 && NbRead != 0)
{
previous_nb_distinct_kmers = nb_distinct_kmers;
nb_distinct_kmers = linearCounter->count()*estimated_nb_reads/NbRead;
//printf("estimated now: %ld\n",nb_distinct_kmers);
// the following condition will grossly over-estimate the number of distinct kmers
// I expect the correct result to be in the same order of magnitude
// 5% error
if (abs((int)(nb_distinct_kmers-previous_nb_distinct_kmers)) < previous_nb_distinct_kmers/20)
stop = true;
if (!linearCounter->is_accurate())
stop = true;
}
}
NbRead++;
if ((NbRead%10000)==0) fprintf (stderr,(char*)"%cExtrapolating number of distinct kmers %lld",13,NbRead);
}
if (!linearCounter->is_accurate())
{
printf("Inaccurate estimation, restarting with %d MB RAM\n",(2*nbytes_memory)/1024/1024);
delete linearCounter;
return extrapolate_distinct_kmers_wrapped(2*nbytes_memory, Reads);
}
nb_distinct_kmers = linearCounter->count()*estimated_nb_reads/NbRead; // this is a very rough estimation
printf("Linear estimation: ~%ld M distinct kmers are in the reads\n",nb_distinct_kmers/1000000L);
delete linearCounter;
return nb_distinct_kmers;
}
