xxay/hifiasm
1
0
Fork 0
mirror of https://github.com/chhylp123/hifiasm.git synced 2026-09-26 08:08:13 +08:00
Code Issues Packages Projects Releases Wiki Activity

prefilter

Browse Source
This commit is contained in:
Haoyu Cheng
2019-07-08 01:06:36 -04:00
parent 87873d6323
commit 1892b4d2db
6 changed files with 1315 additions and 940 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Assembly.cpp
-15
View File
@@ -721,22 +721,7 @@ void* Overlap_calculate_heap_merge(void* arg)
for (i = thr_ID; i < R_INF.total_reads; i = i + thread_num)
{
/**
if (i < 4)
{
continue;
}
if (i > 4)
{
break;
}
fprintf(stderr, "i: %u\n", i);
fflush(stderr);
**/
clear_Heap(&heap);
clear_Candidates_list(&l);
///clear_Candidates_list(&debug_l);
Correct.cpp
+212 -184
View File
@@ -1,6 +1,7 @@
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <pthread.h>
#include "Correct.h"
#include "Levenshtein_distance.h"
#include "edlib.h"
@@ -8,6 +9,7 @@
long long T_total_match=0;
long long T_total_unmatch=0;
long long T_total_mis=0;
pthread_mutex_t debug_statistics ;
#define MAX(x, y) ((x >= y)?x:y)
@@ -286,7 +288,6 @@ long long o_len, int threashold, int error, long long* total_mis)
if (result.status == EDLIB_STATUS_OK) {
if (result.editDistance != error)
///if (result.editDistance != error && result.endLocations[0] > x_len)
{
(*total_mis)++;
@@ -310,36 +311,61 @@ long long o_len, int threashold, int error, long long* total_mis)
char* cigar = edlibAlignmentToCigar(result.alignment, result.alignmentLength, EDLIB_CIGAR_STANDARD);
int cigar_length = strlen(cigar);
/**
for (int i = cigar_length - 1; i >= 0; i--)
int i = cigar_length - 1;
int j = 0;
char tmp;
while (i >= 0)
{
switch (cigar[i])
if (up_length < 0 || left_length < 0)
{
case 'I':
break;
default:
break;
}
}
**/
///fprintf(stderr,"%s\n", cigar);
free(cigar);
/**
fprintf(stderr, "****\ni: %u, edlib: %d, alignmentLength: %d, startLocations: %d, endLocations: %d\n",
i, result.editDistance, result.alignmentLength, result.startLocations[0], result.endLocations[0]);
char* cigar = edlibAlignmentToCigar(result.alignment, result.alignmentLength, EDLIB_CIGAR_STANDARD);
fprintf(stderr,"%s\n", cigar);
free(cigar);
print_string(x_string, x_len);
print_string(y_string, o_len);
fprintf(stderr, "BPM: %d\n", error);
**/
if (cigar[i] == 'I' || cigar[i] == 'D')
{
tmp = cigar[i];
cigar[i] = '\0';
j = i - 1;
while (cigar[j] <= '9' && cigar[j] >= '0' && j >= 0)
{
j--;
}
j++;
int Len = atoi(cigar + j);
cigar[i] = tmp;
i = j - 1;
if (tmp == 'I')
{
up_length = up_length - Len;
left_length = left_length + Len;
}
else
{
left_length = left_length - Len;
up_length = up_length + Len;
}
}
else
{
i--;
}
}
if (up_length >= 0 && left_length >= 0)
{
fprintf(stderr, "****\nedlib: %d, alignmentLength: %d, startLocations: %d, endLocations: %d\n",
result.editDistance, result.alignmentLength, result.startLocations[0], result.endLocations[0]);
fprintf(stderr,"%s\n", cigar);
print_string(x_string, x_len);
print_string(y_string, o_len);
fprintf(stderr, "BPM: %d\n", error);
}
free(cigar);
}
}
edlibFreeAlignResult(result);
@@ -358,10 +384,14 @@ char* r_string)
long long x_end, x_len;
int end_site;
unsigned int error;
long long total_match=0;
long long total_unmatch=0;
long long total_mis=0;
/************需要注释掉********* */
// long long total_match=0;
// long long total_unmatch=0;
// long long total_mis=0;
/************需要注释掉********* */
int groupLen = 0;
int return_sites[GROUP_SIZE];
unsigned int return_sites_error[GROUP_SIZE];
///这些是整个window被完全覆盖的
for (i = 0; i < dumy->length; i++)
@@ -383,54 +413,136 @@ char* r_string)
///不能超过y的剩余长度
o_len = MIN(Window_Len, currentIDLen - y_start);
recover_UC_Read_sub_region(dumy->overlap_region, y_start, o_len, overlap_list->list[currentID].y_pos_strand,
R_INF, overlap_list->list[currentID].y_id);
x_string = r_string + x_start;
y_string = dumy->overlap_region;
///这个长度足够，可以用来一起比
if (Window_Len == o_len)
{
end_site = Reserve_Banded_BPM(y_string, o_len, x_string, x_len, THRESHOLD, &error);
recover_UC_Read_sub_region(dumy->overlap_region_group[groupLen], y_start, o_len, overlap_list->list[currentID].y_pos_strand,
R_INF, overlap_list->list[currentID].y_id);
if (error!=(unsigned int)-1)
x_string = r_string + x_start;
groupLen++;
if (groupLen == GROUP_SIZE)
{
total_match++;
}
else
{
total_unmatch++;
}
test_edit_distance_by_edlib(x_string, y_string, x_len,
o_len, THRESHOLD, error, &total_mis);
Reserve_Banded_BPM_4_SSE_only(dumy->overlap_region_group[0], dumy->overlap_region_group[1],
dumy->overlap_region_group[2], dumy->overlap_region_group[3], o_len, x_string, x_len,
return_sites, return_sites_error, THRESHOLD, dumy->Peq_SSE);
groupLen = 0;
// end_site = Reserve_Banded_BPM(dumy->overlap_region_group[0], o_len, x_string, x_len, THRESHOLD, &error);
// if (error!=return_sites_error[0])
// {
// fprintf(stderr, "error\n");
// }
// end_site = Reserve_Banded_BPM(dumy->overlap_region_group[1], o_len, x_string, x_len, THRESHOLD, &error);
// if (error!=return_sites_error[1])
// {
// fprintf(stderr, "error\n");
// }
// end_site = Reserve_Banded_BPM(dumy->overlap_region_group[2], o_len, x_string, x_len, THRESHOLD, &error);
// if (error!=return_sites_error[2])
// {
// fprintf(stderr, "error\n");
// }
// end_site = Reserve_Banded_BPM(dumy->overlap_region_group[3], o_len, x_string, x_len, THRESHOLD, &error);
// if (error!=return_sites_error[3])
// {
// fprintf(stderr, "error\n");
// }
// /************需要注释掉**********/
// for (size_t ijk = 0; ijk < GROUP_SIZE; ijk++)
// {
// if (return_sites_error[ijk]!=(unsigned int)-1)
// {
// total_match++;
// }
// else
// {
// total_unmatch++;
// }
// test_edit_distance_by_edlib(x_string, dumy->overlap_region_group[ijk], x_len, o_len, THRESHOLD,
// return_sites_error[ijk], &total_mis);
// }
// /************需要注释掉********* */
}
}
else ///这个不够，只能单个比 ///不够的地方要置N
{
/**
recover_UC_Read_sub_region(dumy->overlap_region, y_start, o_len, overlap_list->list[currentID].y_pos_strand,
R_INF, overlap_list->list[currentID].y_id);
x_string = r_string + x_start;
y_string = dumy->overlap_region;
///o_len < Window_Len, 说明y的长度不够，需要在y后面补N
memset (y_string + o_len, 0, Window_Len - o_len);
end_site = Reserve_Banded_BPM(y_string, o_len, x_string, x_len, THRESHOLD, &error);
if (error!=-1)
{
total_match++;
}
else
{
total_unmatch++;
}
**/
// /************需要注释掉**********/
// if (error!=(unsigned int)-1)
// {
// total_match++;
// }
// else
// {
// total_unmatch++;
// }
// test_edit_distance_by_edlib(x_string, y_string, x_len, o_len, THRESHOLD, error, &total_mis);
// /************需要注释掉********* */
}
}
if (groupLen == 1)
{
end_site = Reserve_Banded_BPM(dumy->overlap_region_group[0], o_len, x_string, x_len, THRESHOLD, &error);
// /************需要注释掉**********/
// if (error!=(unsigned int)-1)
// {
// total_match++;
// }
// else
// {
// total_unmatch++;
// }
// test_edit_distance_by_edlib(x_string, dumy->overlap_region_group[0], x_len, o_len, THRESHOLD, error, &total_mis);
// /************需要注释掉********* */
}
else if (groupLen > 1)
{
Reserve_Banded_BPM_4_SSE_only(dumy->overlap_region_group[0], dumy->overlap_region_group[1],
dumy->overlap_region_group[2], dumy->overlap_region_group[3], o_len, x_string, x_len,
return_sites, return_sites_error, THRESHOLD, dumy->Peq_SSE);
// /************需要注释掉**********/
// for (size_t ijk = 0; ijk < groupLen; ijk++)
// {
// if (return_sites_error[ijk]!=(unsigned int)-1)
// {
// total_match++;
// }
// else
// {
// total_unmatch++;
// }
// test_edit_distance_by_edlib(x_string, dumy->overlap_region_group[ijk], x_len, o_len, THRESHOLD,
// return_sites_error[ijk], &total_mis);
// }
// /************需要注释掉**********/
groupLen = 0;
}
long long reverse_i = dumy->size - 1;
///这些是整个window被部分覆盖的
@@ -471,41 +583,38 @@ char* r_string)
y_string = dumy->overlap_region;
///不够的地方要置N
if (Window_Len != o_len)
{
///o_len < Window_Len, 说明y的长度不够，需要在y后面补N
memset (y_string + o_len, 0, Window_Len - o_len);
}
end_site = Reserve_Banded_BPM(y_string, o_len, x_string, x_len, THRESHOLD, &error);
/**
if (error!=-1)
{
total_match++;
}
else
{
total_unmatch++;
}
**/
// /************需要注释掉********* */
// if (error!=(unsigned int)-1)
// {
// total_match++;
// }
// else
// {
// total_unmatch++;
// }
// test_edit_distance_by_edlib(x_string, y_string, x_len, o_len, THRESHOLD, error, &total_mis);
// /************需要注释掉********* */
}
/**
if (total_match!=0)
{
// /************需要注释掉********* */
// pthread_mutex_lock(&debug_statistics);
// T_total_match = T_total_match + total_match;
// T_total_unmatch = T_total_unmatch + total_unmatch;
// T_total_mis = T_total_mis + total_mis;
// pthread_mutex_unlock(&debug_statistics);
// /************需要注释掉********* */
fprintf(stderr, "total_match: %u\n", total_match);
fprintf(stderr, "total_unmatch: %u\n", total_unmatch);
fprintf(stderr, "total_mis: %u\n", total_mis);
}
**/
T_total_match = T_total_match + total_match;
T_total_unmatch = T_total_unmatch + total_unmatch;
T_total_mis = T_total_mis + total_mis;
}
void correct_overlap(overlap_region_alloc* overlap_list, All_reads* R_INF, UC_Read* g_read, Correct_dumy* dumy)
@@ -522,90 +631,6 @@ void correct_overlap(overlap_region_alloc* overlap_list, All_reads* R_INF, UC_Re
window_start = 0;
window_end = WINDOW - 1;
/**
UC_Read debug_read;
init_UC_Read(&debug_read);
for (i = 0; i < overlap_list->length; i++)
{
if (overlap_list->list[i].y_pos_strand)
{
recover_UC_Read_RC(&debug_read, R_INF, overlap_list->list[i].y_id);
}
else
{
recover_UC_Read(&debug_read, R_INF, overlap_list->list[i].y_id);
}
long long x_length = overlap_list->list[i].x_pos_e - overlap_list->list[i].x_pos_s + 1;
long long y_length = overlap_list->list[i].y_pos_e - overlap_list->list[i].y_pos_s + 1;
EdlibAlignResult result = edlibAlign(g_read->seq+overlap_list->list[i].x_pos_s,
x_length,
debug_read.seq + overlap_list->list[i].y_pos_s,
y_length,
edlibNewAlignConfig(-1, EDLIB_MODE_NW, EDLIB_TASK_DISTANCE, NULL, 0));
if (result.status == EDLIB_STATUS_OK) {
if (result.editDistance>0 && result.editDistance < x_length*0.02)
{
fprintf(stderr, "inner_i: %u, x_length: %u, y_length: %u, editDistance: %u\n",
i, x_length, y_length, result.editDistance);
fprintf(stderr, "x_pos_s: %u, x_pos_e: %u\n",
overlap_list->list[i].x_pos_s, overlap_list->list[i].x_pos_e);
fprintf(stderr, "y_pos_s: %u, y_pos_e: %u, y_pos_strand: %u\n",
overlap_list->list[i].y_pos_s, overlap_list->list[i].y_pos_e, overlap_list->list[i].y_pos_strand);
EdlibAlignResult n_result = edlibAlign(g_read->seq+overlap_list->list[i].x_pos_s,
350,
debug_read.seq + overlap_list->list[i].y_pos_s - 15,
380,
edlibNewAlignConfig(-1, EDLIB_MODE_HW, EDLIB_TASK_DISTANCE, NULL, 0));
fprintf(stderr, "n_editDistance: %u\n",
n_result.editDistance);
edlibFreeAlignResult(n_result);
}
}
edlibFreeAlignResult(result);
}
destory_UC_Read(&debug_read);
**/
///return;
int flag;
for (i = 0; i < window_num; i++)
@@ -641,18 +666,16 @@ void correct_overlap(overlap_region_alloc* overlap_list, All_reads* R_INF, UC_Re
{
window_end = g_read->length - 1;
}
///break;
}
fprintf(stderr, "total_match: %u, total_unmatch: %u, total_mis: %u\n",
T_total_match, T_total_unmatch, T_total_mis);
// /************需要注释掉********* */
// pthread_mutex_lock(&debug_statistics);
// fprintf(stderr, "total_match: %u, total_unmatch: %u, total_mis: %u\n",
// T_total_match, T_total_unmatch, T_total_mis);
// pthread_mutex_unlock(&debug_statistics);
// /************需要注释掉********* */
}
@@ -663,6 +686,11 @@ void init_Correct_dumy(Correct_dumy* list)
list->lengthNT = 0;
list->start_i = 0;
list->overlapID = NULL;
int i;
for (i = 0; i < 256; i++)
{
list->Peq_SSE[i] = _mm_setzero_si128();
}
}
void destory_Correct_dumy(Correct_dumy* list)
Correct.h
+4
View File
@@ -2,9 +2,11 @@
#define __CORRECT__
#include <stdint.h>
#include "Hash_Table.h"
#include "Levenshtein_distance.h"
#define WINDOW 350
#define THRESHOLD 15
#define GROUP_SIZE 4
typedef struct
{
@@ -14,6 +16,8 @@ typedef struct
uint64_t size;
uint64_t start_i;
char overlap_region[WINDOW + THRESHOLD*2 + 10];
char overlap_region_group[GROUP_SIZE][WINDOW + THRESHOLD*2 + 10];
__m128i Peq_SSE[256];
} Correct_dumy;
Hash_Table.cpp
+14
View File
@@ -3,6 +3,7 @@
#include <stdint.h>
#include "Hash_Table.h"
#include "Process_Read.h"
#include "Correct.h"
#include <pthread.h>
pthread_mutex_t output_mutex;
@@ -715,6 +716,19 @@ inline void append_pos_to_Candidates_list(Candidates_list* candidates, ElemType*
candidates->length++;
}
void pre_filter_by_nearby(k_mer_pos* n_list, uint64_t n_length, uint64_t n_end_pos, uint8_t n_direction, UC_Read* g_read)
{
char* x_string = NULL;
char* y_string = NULL;
long long i = 0;
for (i = 0; i < n_length; i++)
{
/* code */
}
}
void test_single_list(Candidates_list* candidates, k_mer_pos* n_list, uint64_t n_lengh, uint64_t end_pos, uint64_t strand)
{
Levenshtein_distance.cpp
+805
View File
@@ -1 +1,806 @@
#include "Levenshtein_distance.h"
void output_bit_myers(Word x, int length)
{
int i = 0;
while (i < length)
{
fprintf(stderr, "%u", (x >> i) & ((Word)1));
i++;
}
fprintf(stderr, "\n");
}
void prase_vertical(Word VP, Word VN, int length, int matrix[1000][1000], int i)
{
i++;
int k = 0;
int j = i;
int diff;
while (k < length)
{
int x_p = (VP >> k) & ((Word)1);
int x_n = (VN >> k) & ((Word)1);
if (x_p == 1 && x_n == 1)
{
fprintf(stderr, "error\n");
}
if (x_p == 1)
{
diff = 1;
///fprintf(stderr, "[+1]");
}
if (x_n == 1)
{
diff = -1;
///fprintf(stderr, "[-1]");
}
if (x_p == 0 && x_n == 0)
{
diff = 0;
///fprintf(stderr, "[+0]");
}
j++;
if (matrix[i][j] - matrix[i][j - 1] != diff)
{
fprintf(stderr, "*************V(k): %u\n", k);
///return;
}
k++;
}
///fprintf(stderr, "\n");
}
void prase_D0(Word D0, int length, int matrix[1000][1000], int i)
{
i++;
int k = 0;
int j = i;
int diff;
while (k < length)
{
int diff = (D0 >> k) & ((Word)1);
if(diff==matrix[i][j] - matrix[i-1][j-1])
{
fprintf(stderr, "*************D(k): %u\n", k);
fprintf(stderr, "diff: %u, matrix[i][j]: %u, matrix[i-1][j-1]: %u\n", diff, matrix[i][j], matrix[i-1][j-1]);
///return;
}
j++;
k++;
}
///fprintf(stderr, "\n");
}
void prase_H(Word HP, Word HN, int length, int matrix[1000][1000], int i)
{
i++;
int k = 0;
int j = i;
int diff;
while (k < length)
{
int x_p = (HP >> k) & ((Word)1);
int x_n = (HN >> k) & ((Word)1);
if (x_p == 1 && x_n == 1)
{
fprintf(stderr, "error\n");
}
if (x_p == 1)
{
diff = 1;
///fprintf(stderr, "[+1]");
}
if (x_n == 1)
{
diff = -1;
///fprintf(stderr, "[-1]");
}
if (x_p == 0 && x_n == 0)
{
diff = 0;
///fprintf(stderr, "[+0]");
}
if(diff!=matrix[i][j] - matrix[i-1][j])
{
fprintf(stderr, "*************H(k): %u\n", k);
///return;
}
j++;
k++;
}
///fprintf(stderr, "\n");
}
/**
pattern是长的那个，是y
p_length是长的那个的长度, p_length实际没用
text是短的那个，是x
t_length是短的那个的长度
errthold是阈值
return_err是编辑距离
返回值是结束位置
**/
int Reserve_Banded_BPM_debug
(char *pattern, int p_length, char *text, int t_length, unsigned short errthold, unsigned int* return_err, int matrix[1000][1000])
{
(*return_err) = (unsigned int)-1;
Word Peq[256];
int band_length = (errthold << 1) + 1;
int i = 0;
Word tmp_Peq_1 = (Word)1;
Peq['A'] = (Word)0;
Peq['T'] = (Word)0;
Peq['G'] = (Word)0;
Peq['C'] = (Word)0;
Word Peq_A;
Word Peq_T;
Word Peq_C;
Word Peq_G;
///band_length = 2k + 1
for (i = 0; i<band_length; i++)
{
Peq[pattern[i]] = Peq[pattern[i]] | tmp_Peq_1;
tmp_Peq_1 = tmp_Peq_1 << 1;
}
///Peq['T'] = Peq['T'] | Peq['C'];
Peq_A = Peq['A'];
Peq_C = Peq['C'];
Peq_T = Peq['T'];
Peq_G = Peq['G'];
memset(Peq, 0, sizeof(Word)* 256);
Peq['A'] = Peq_A;
Peq['C'] = Peq_C;
Peq['T'] = Peq_T;
Peq['G'] = Peq_G;
Word Mask = ((Word)1 << (errthold << 1));
Word VP = 0;
Word VN = 0;
Word X = 0;
Word D0 = 0;
Word HN = 0;
Word HP = 0;
i = 0;
int err = 0;
Word err_mask = (Word)1;
///band_down = 2k
///i_bd = 2k
///int i_bd = i + band_down;
int i_bd = (errthold << 1);
int last_high = (errthold << 1);
/// t_length_1 = SEQ_LENGTH - 1
int t_length_1 = t_length - 1;
//while(i<t_length)
while (i<t_length_1)
{
fprintf(stderr, "i: %u, j_s: %u, j_e: %u\n", i + 1, i + 1, i + band_length + 1);
///if (i >= 6)
if (i >= 0)
{
/**
fprintf(stderr, "VP:\n");
output_bit_myers(VP, band_length);
fprintf(stderr, "VN:\n");
output_bit_myers(VN, band_length);
fprintf(stderr, "HP:\n");
output_bit_myers(HP, band_length);
fprintf(stderr, "HN:\n");
output_bit_myers(HN, band_length);
fprintf(stderr, "D0:\n");
output_bit_myers(D0, band_length);
fprintf(stderr, "text[i]: %c\n", text[i]);
fprintf(stderr, "Peq[text[i]]:\n");
output_bit_myers(Peq[text[i]], band_length);
for (size_t j = i; j < i + band_length; j++)
{
fprintf(stderr, "%c", pattern[j]);
}
fprintf(stderr, "\n");
fprintf(stderr, "Previous begin.\n");
prase_vertical(VP, VN, band_length, matrix, i-1);
prase_D0(D0, band_length, matrix, i-1);
prase_H(HP, HN, band_length, matrix, i-1);
fprintf(stderr, "Previous test done.\n");
**/
X = Peq[text[i]] | VN;
/**
fprintf(stderr, "#X:\n");
output_bit_myers(X, band_length);
**/
D0 = ((VP + (X&VP)) ^ VP) | X;
/**
fprintf(stderr, "#(X&VP):\n");
output_bit_myers((X&VP), band_length);
fprintf(stderr, "#(VP + (X&VP)):\n");
output_bit_myers((VP + (X&VP)), band_length);
fprintf(stderr, "#((VP + (X&VP)) ^ VP):\n");
output_bit_myers(((VP + (X&VP)) ^ VP), band_length);
fprintf(stderr, "#D0:\n");
output_bit_myers(D0, band_length);
**/
HN = VP&D0;
HP = VN | ~(VP | D0);
X = D0 >> 1;
VN = X&HP;
VP = HN | ~(X | HP);
}
else
{
///pattern[0]ÔÚPeq[2k], ¶øpattern[2k]ÔÚPeq[0]
X = Peq[text[i]] | VN;
D0 = ((VP + (X&VP)) ^ VP) | X;
HN = VP&D0;
HP = VN | ~(VP | D0);
X = D0 >> 1;
VN = X&HP;
VP = HN | ~(X | HP);
}
/**
for (size_t j = i + 1; j <= i + band_length + 1; j++)
{
fprintf(stderr, "[%u]", matrix[i + 1][j]);
}
fprintf(stderr, "\n");
**/
prase_vertical(VP, VN, band_length, matrix, i);
prase_D0(D0, band_length, matrix, i);
prase_H(HP, HN, band_length, matrix, i);
/**
fprintf(stderr, "VP:\n");
output_bit_myers(VP, band_length);
fprintf(stderr, "VN:\n");
output_bit_myers(VN, band_length);
**/
if (!(D0&err_mask))
{
++err;
///¼´Ê¹È«²¿µÝ¼õ£¬Ò²¾Í¼õ2k
if ((err - last_high)>errthold)
{
///fprintf(stderr, "0 ######, i: %u\n", i);
return -1;
}
}
Peq['A'] = Peq['A'] >> 1;
Peq['C'] = Peq['C'] >> 1;
Peq['G'] = Peq['G'] >> 1;
Peq['T'] = Peq['T'] >> 1;
++i;
++i_bd;
Peq[pattern[i_bd]] = Peq[pattern[i_bd]] | Mask;
///Peq['T'] = Peq['T'] | Peq['C'];
}
X = Peq[text[i]] | VN;
D0 = ((VP + (X&VP)) ^ VP) | X;
HN = VP&D0;
HP = VN | ~(VP | D0);
X = D0 >> 1;
VN = X&HP;
VP = HN | ~(X | HP);
if (!(D0&err_mask))
{
++err;
if ((err - last_high)>errthold)
return -1;
}
prase_vertical(VP, VN, band_length, matrix, i);
prase_D0(D0, band_length, matrix, i);
prase_H(HP, HN, band_length, matrix, i);
fprintf(stderr, "err: %d, matrix[][]: %d\n", err, matrix[i+1][i+1]);
fprintf(stderr, "VP:\n");
output_bit_myers(VP, band_length);
fprintf(stderr, "VN:\n");
output_bit_myers(VN, band_length);
////fprintf(stderr, "sucess(2)\n");
/// last_high = 2k
/// site = (SEQ_LENGTH + 2k) - 2k -1
/// site = SEQ_LENGTH - 1
///int site = p_length - last_high - 1;
int site = t_length - 1;
int return_site = -1;
if ((err <= errthold) && (err<=*return_err))
{
*return_err = err;
return_site = site;
}
int i_last = i;
i = 0;
while (i<errthold)
{
err = err + ((VP >> i)&(Word)1);
err = err - ((VN >> i)&(Word)1);
++i;
fprintf(stderr, "*i: %u, err: %d\n", i, err);
if ((err <= errthold) && (err <= *return_err))
{
*return_err = err;
return_site = site + i;
}
}
unsigned int ungap_err;
ungap_err = err;
while (i<last_high)
{
err = err + ((VP >> i)&(Word)1);
err = err - ((VN >> i)&(Word)1);
++i;
fprintf(stderr, "*i: %u, err: %d\n", i, err);
if ((err <= errthold) && (err<=*return_err))
{
*return_err = err;
return_site = site + i;
}
}
if ((ungap_err <= errthold) && (ungap_err == *return_err))
{
return_site = site + errthold;
}
return return_site;
}
int BS_Reserve_Banded_BPM
(char *pattern, int p_length, char *text, int t_length, unsigned short errthold, unsigned int* return_err)
{
(*return_err) = (unsigned int)-1;
///Õâ¸öÊÇÄǸöÐèÒªÔ¤´¦ÀíµÄÏòÁ¿
Word Peq[256];
int band_length = (errthold << 1) + 1;
int i = 0;
Word tmp_Peq_1 = (Word)1;
Peq['A'] = (Word)0;
Peq['T'] = (Word)0;
Peq['G'] = (Word)0;
Peq['C'] = (Word)0;
Word Peq_A;
Word Peq_T;
Word Peq_C;
Word Peq_G;
///band_length = 2k + 1
///ÕâÊÇ°ÑpatternµÄÇ°2k + 1¸ö×Ö·ûÔ¤´¦Àí
///pattern[0]¶ÔÓ¦Peq[0]
///pattern[2k]¶ÔÓ¦Peq[2k]
for (i = 0; i<band_length; i++)
{
Peq[pattern[i]] = Peq[pattern[i]] | tmp_Peq_1;
tmp_Peq_1 = tmp_Peq_1 << 1;
}
Peq['T'] = Peq['T'] | Peq['C'];
Peq_A = Peq['A'];
Peq_C = Peq['C'];
Peq_T = Peq['T'];
Peq_G = Peq['G'];
memset(Peq, 0, sizeof(Word)* 256);
Peq['A'] = Peq_A;
Peq['C'] = Peq_C;
Peq['T'] = Peq_T;
Peq['G'] = Peq_G;
Word Mask = ((Word)1 << (errthold << 1));
Word VP = 0;
Word VN = 0;
Word X = 0;
Word D0 = 0;
Word HN = 0;
Word HP = 0;
i = 0;
int err = 0;
Word err_mask = (Word)1;
///band_down = 2k
///i_bd = 2k
///int i_bd = i + band_down;
int i_bd = (errthold << 1);
int last_high = (errthold << 1);
/// t_length_1 = SEQ_LENGTH - 1
int t_length_1 = t_length - 1;
//while(i<t_length)
while (i<t_length_1)
{
///pattern[0]ÔÚPeq[2k], ¶øpattern[2k]ÔÚPeq[0]
X = Peq[text[i]] | VN;
D0 = ((VP + (X&VP)) ^ VP) | X;
HN = VP&D0;
HP = VN | ~(VP | D0);
X = D0 >> 1;
VN = X&HP;
VP = HN | ~(X | HP);
///Èç¹ûб¶Ô½ÇÏß·½ÏòÆ¥ÅäÔòD0ÊÇ1
///Èç¹û²»Æ¥ÅäÔòD0ÊÇ0
///Õâ¸öÒâ˼ÊÇÈç¹û×îÉÏÃæÄÇÌõ¶Ô½ÇÏßÉϵÄб¶Ô½ÇÏß·½Ïò·¢ÉúÎóÅä,ÔòÖ´ÐÐÄÚ²¿³ÌÐò
///
if (!(D0&err_mask))
{
++err;
///¼´Ê¹È«²¿µÝ¼õ£¬Ò²¾Í¼õ2k
if ((err - last_high)>errthold)
return -1;
}
///pattern[0]ÔÚPeq[2k], ¶øpattern[2k]ÔÚPeq[0]
//ÓÒÒÆʵ¼ÊÉÏÊÇ°Ñpattern[0]ÒƵôÁË
Peq['A'] = Peq['A'] >> 1;
Peq['C'] = Peq['C'] >> 1;
Peq['G'] = Peq['G'] >> 1;
Peq['T'] = Peq['T'] >> 1;
++i;
++i_bd;
///ÕâÊÇ°ÑеÄpattern[2k]¼Ó½øÀ´, ÕâòËÆÊǼӵ½Peq[2k]ÉÏÁË
Peq[pattern[i_bd]] = Peq[pattern[i_bd]] | Mask;
Peq['T'] = Peq['T'] | Peq['C'];
}
///fprintf(stderr, "sucess(1)\n");
///Õâ¸öÑ­»·ÄóöÀ´ÊÇΪÁË·ÀÖ¹ÄÚ´æй¶
///ÆäʵҲ¾ÍÊÇÑ­»·ÀïµÄ×îºóÒ»ÐÐÓï¾ä°É
///ÍêÈ«¿ÉÒÔ°ÑpatternÔö´óһλ
///²»¹ýÕâÑùÒ²ºÃ£¬¿ÉÒÔ¼õÉÙ¼ÆË㿪Ïú
X = Peq[text[i]] | VN;
D0 = ((VP + (X&VP)) ^ VP) | X;
HN = VP&D0;
HP = VN | ~(VP | D0);
X = D0 >> 1;
VN = X&HP;
VP = HN | ~(X | HP);
if (!(D0&err_mask))
{
++err;
if ((err - last_high)>errthold)
return -1;
}
////fprintf(stderr, "sucess(2)\n");
/// last_high = 2k
/// site = (SEQ_LENGTH + 2k) - 2k -1
/// site = SEQ_LENGTH - 1
///´ËʱÕâ¸ösiteòËÆÊÇ×îÉÏÃæÄÇÌõ¶Ô½ÇÏßµÄλÖÃ
///int site = p_length - last_high - 1;
int site = t_length - 1;
int return_site = -1;
if ((err <= errthold) && (err<=*return_err))
{
*return_err = err;
return_site = site;
}
int i_last = i;
i = 0;
while (i<errthold)
{
err = err + ((VP >> i)&(Word)1);
err = err - ((VN >> i)&(Word)1);
++i;
if ((err <= errthold) && (err <= *return_err))
{
*return_err = err;
return_site = site + i;
}
}
unsigned int ungap_err;
ungap_err = err;
while (i<last_high)
{
err = err + ((VP >> i)&(Word)1);
err = err - ((VN >> i)&(Word)1);
++i;
if ((err <= errthold) && (err<=*return_err))
{
*return_err = err;
return_site = site + i;
}
}
if ((ungap_err <= errthold) && (ungap_err == *return_err))
{
return_site = site + errthold;
}
return return_site;
}
int Reserve_Banded_BPM_new(char *pattern,int p_length,char *text,int t_length,
unsigned short errthold,unsigned short band_down,unsigned short band_below,unsigned short band_length,int* return_err, int thread_id)
{
Word Peq[128];
char Peq_index[4]= {'A','C','G','T'};
int symbol = 0;
int r;
Word tmp_Peq_1=(Word)1;
Peq['A']=(Word)0;
Peq['T']=(Word)0;
Peq['G']=(Word)0;
Peq['C']=(Word)0;
Word Peq_A;
Word Peq_T;
Word Peq_C;
Word Peq_G;
for (r =0; r<band_length; r++)
{
Peq[pattern[r]]=Peq[pattern[r]]|tmp_Peq_1;
tmp_Peq_1=tmp_Peq_1<<1;
}
Peq_A=Peq['A'];
Peq_C=Peq['C'];
Peq_T=Peq['T'];
Peq_G=Peq['G'];
for(symbol = 0; symbol < 128; symbol++)
{
Peq[symbol]=(Word)0;
//jump_c[symbol]=128;
}
Peq['A']=Peq_A;
Peq['C']=Peq_C;
Peq['T']=Peq_T;
Peq['G']=Peq_G;
Word Mask_Pre=(Word)1<<(band_length-2);
Word Mask=(Word)1<<(band_length-1);
Word VP=0;
Word VN=0;
Word X=0;
Word D0=0;
Word HN=0;
Word HP=0;
int s=0;
int i = 0;
int j=0;
int bound=band_length-2-band_down;
int err=0;
Word err_mask=(Word)1;
int s1=band_length-2;
int i_bd=i+band_down;
int last_high=band_length-t_length+p_length-band_down-1;
int t_length_1=t_length-1;
//while(i<t_length)
while(i<t_length_1)
{
X=Peq[text[i]]|VN;
D0=((VP+(X&VP))^VP)|X;
HN=VP&D0;
HP=VN|~(VP|D0);
X=D0>>1;
VN=X&HP;
VP=HN|~(X|HP);
if(!(D0&err_mask))
{
++err;
if((err-last_high)>errthold)
return -1;
}
Peq['A']=Peq['A']>>1;
Peq['C']=Peq['C']>>1;
Peq['G']=Peq['G']>>1;
Peq['T']=Peq['T']>>1;
++i;
++i_bd;
Peq[pattern[i_bd]]=Peq[pattern[i_bd]]|Mask;
}
///这个循环拿出来是为了防止内存泄露
X=Peq[text[i]]|VN;
D0=((VP+(X&VP))^VP)|X;
HN=VP&D0;
HP=VN|~(VP|D0);
X=D0>>1;
VN=X&HP;
VP=HN|~(X|HP);
if(!(D0&err_mask))
{
++err;
if((err-last_high)>errthold)
return -1;
}
int site=p_length-last_high-1;
int return_site=-1;
if((err<=errthold)&&(err<*return_err))
{
*return_err=err;
return_site=site;
}
int i_last=i;
i=0;
while(i<last_high)
{
err=err+((VP>>i)&(Word)1);
err=err-((VN>>i)&(Word)1);
++i;
if((err<=errthold)&&(err<*return_err))
{
*return_err=err;
return_site=site+i;
}
}
return return_site;
}
Levenshtein_distance.h
+280 -741
View File
File diff suppressed because it is too large Load Diff