zzh
/
bwa_perf
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

解决了各种bug，目前mid interval和普通的occ interval合并在一起处理了，通过编译指导语句，进行选择和调试

This commit is contained in:
zzh 2024-02-02 22:18:27 +08:00
parent 5ff04ec853
commit 45318fc95b
4 changed files with 137 additions and 236 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
.vscode/launch.json vendored
View File

@ -10,7 +10,7 @@
"request": "launch", "request": "launch",
"program": "${workspaceRoot}/fmtidx", "program": "${workspaceRoot}/fmtidx",
"args": [ "args": [
"/mnt/d/data/reference/human_g1k_v37_decoy.fasta" "/home/zzh/reference/human_g1k_v37_decoy.fasta"
], ],
"cwd": "${workspaceFolder}", // 当前工作路径：当前文件所在的工作空间 "cwd": "${workspaceFolder}", // 当前工作路径：当前文件所在的工作空间
} }

2
bwt.cpp
View File

@ -354,7 +354,7 @@ bwtintv_t bwt_search2(bwt_t *bwt, const string &q)
bwt_set_intv(bwt, bval(q[x]), ik); bwt_set_intv(bwt, bval(q[x]), ik);
ik.info = x + 1; ik.info = x + 1;
// cout << "bwt-2: " << ik.x[0] << '\t' << ik.x[1] << '\t' << ik.x[2] << endl; //cout << "bwt-2: " << ik.x[0] << '\t' << ik.x[1] << '\t' << ik.x[2] << endl;
for (i = x + 1; i + 1 < (int)q.size(); i += 2) for (i = x + 1; i + 1 < (int)q.size(); i += 2)
{ {

351
fmt_index.cpp
View File

@ -8,6 +8,7 @@
#include <algorithm> #include <algorithm>
#include <string.h> #include <string.h>
#include <immintrin.h> #include <immintrin.h>
#include <sstream>
#include "util.h" #include "util.h"
#include "bwt.h" #include "bwt.h"
@ -203,12 +204,16 @@ FMTIndex *create_fmt_from_bwt(bwt_t *bwt)
fmt->bwt_size += n_occ * 20; // A,C,G,T和AA,AC.....TG,TT共20个 fmt->bwt_size += n_occ * 20; // A,C,G,T和AA,AC.....TG,TT共20个
#ifdef FMT_MID_INTERVAL #ifdef FMT_MID_INTERVAL
uint32_t s1;
bwtint_t mn_occ = (bwt->seq_len >> FMT_OCC_INTV_SHIFT) * (FMT_OCC_INTERVAL / FMT_MID_INTERVAL - 1); bwtint_t mn_occ = (bwt->seq_len >> FMT_OCC_INTV_SHIFT) * (FMT_OCC_INTERVAL / FMT_MID_INTERVAL - 1);
bwtint_t last_seq_len = bwt->seq_len % FMT_OCC_INTERVAL; bwtint_t last_seq_len = bwt->seq_len % FMT_OCC_INTERVAL;
mn_occ += (last_seq_len + FMT_MID_INTERVAL - 1) / FMT_MID_INTERVAL - 1; mn_occ += (last_seq_len + FMT_MID_INTERVAL - 1) / FMT_MID_INTERVAL - 1;
fmt->bwt_size += mn_occ * 4; fmt->bwt_size += mn_occ * 4;
//cout << "mn_occ: " << mn_occ << endl; cout << (FMT_OCC_INTERVAL / FMT_MID_INTERVAL - 1) << ' ' << last_seq_len << ' ' << (last_seq_len + FMT_MID_INTERVAL - 1) / FMT_MID_INTERVAL - 1 << endl;
//exit(0); cout << "mn_occ: " << mn_occ << ' ' << mn_occ * 4 << endl;
i = 0;
//cout << ((i + 16 & FMT_MID_INTV_MASK) == 0) << endl;
// exit(0);
#endif #endif
buf = (uint32_t *)calloc(fmt->bwt_size, 4); // 开辟计算fmt用到的缓存 buf = (uint32_t *)calloc(fmt->bwt_size, 4); // 开辟计算fmt用到的缓存
@ -304,47 +309,35 @@ FMTIndex *create_fmt_from_bwt(bwt_t *bwt)
if (j > 8) if (j > 8)
{ {
#if FMT_MID_INTERVAL == 8
for (m = 0; m < 4; ++m)
{
uint32_t s1 = pre_and_bwt_seq;
mc[m] += cnt_table[m][s1 & 0xff] + cnt_table[m][s1 >> 8 & 0xff] + cnt_table[m][s1 >> 16 & 0xff] + cnt_table[m][s1 >> 24];
buf[k++] = mc[m];
}
#endif
for (m = 8; m > 0; --m) for (m = 8; m > 0; --m)
{ {
int lshift_bit = 2 * m - 2; int lshift_bit = 2 * m - 2;
pre_and_bwt_seq_2 |= (((pre_bwt_16_seq & (3 << lshift_bit)) << (m * 2)) | ((bwt_16_seq & (3 << lshift_bit)) << (m * 2 - 2))); pre_and_bwt_seq_2 |= (((pre_bwt_16_seq & (3 << lshift_bit)) << (m * 2)) | ((bwt_16_seq & (3 << lshift_bit)) << (m * 2 - 2)));
} }
buf[k++] = pre_and_bwt_seq_2;
#ifdef FMT_MID_INTERVAL
s1 = pre_and_bwt_seq;
for (m = 0; m < 4; ++m)
mc[m] += cnt_table[m][s1 & 0xff] + cnt_table[m][s1 >> 8 & 0xff] + cnt_table[m][s1 >> 16 & 0xff] + cnt_table[m][s1 >> 24];
#endif
#if FMT_MID_INTERVAL == 8 #if FMT_MID_INTERVAL == 8
if ((i + 16) % FMT_OCC_INTERVAL != 0 && j == 16)
for (m = 0; m < 4; ++m)
{
uint32_t s1 = pre_and_bwt_seq_2;
mc[m] += cnt_table[m][s1 & 0xff] + cnt_table[m][s1 >> 8 & 0xff] + cnt_table[m][s1 >> 16 & 0xff] + cnt_table[m][s1 >> 24];
buf[k++] = mc[m];
}
#endif
}
#if FMT_MID_INTERVAL == 16
// 加入中间的check point
for (m = 0; m < 4; ++m)
{
uint32_t s1 = pre_and_bwt_seq;
mc[m] += cnt_table[m][s1 & 0xff] + cnt_table[m][s1 >> 8 & 0xff] + cnt_table[m][s1 >> 16 & 0xff] + cnt_table[m][s1 >> 24];
s1 = pre_and_bwt_seq_2;
mc[m] += cnt_table[m][s1 & 0xff] + cnt_table[m][s1 >> 8 & 0xff] + cnt_table[m][s1 >> 16 & 0xff] + cnt_table[m][s1 >> 24];
}
if ((i + 16) % FMT_OCC_INTERVAL != 0 && j > 8) // 序列最后的部分不用保存
{
for (m = 0; m < 4; ++m) for (m = 0; m < 4; ++m)
buf[k++] = mc[m]; buf[k++] = mc[m];
}
#endif #endif
buf[k++] = pre_and_bwt_seq_2;
#ifdef FMT_MID_INTERVAL
s1 = pre_and_bwt_seq_2;
for (m = 0; m < 4; ++m)
mc[m] += cnt_table[m][s1 & 0xff] + cnt_table[m][s1 >> 8 & 0xff] + cnt_table[m][s1 >> 16 & 0xff] + cnt_table[m][s1 >> 24];
if ((i + 16) % FMT_OCC_INTERVAL != 0 && j == 16 && ((i + 16) & FMT_MID_INTV_MASK) == 0)
for (m = 0; m < 4; ++m)
buf[k++] = mc[m];
#endif
}
} }
} }
// the last element // the last element
@ -354,6 +347,7 @@ FMTIndex *create_fmt_from_bwt(bwt_t *bwt)
k += 16; k += 16;
// cout << "n occ: " << n_occ << endl; // cout << "n occ: " << n_occ << endl;
cout << "size: " << k << '\t' << fmt->bwt_size << endl; cout << "size: " << k << '\t' << fmt->bwt_size << endl;
// exit(0);
xassert(k == fmt->bwt_size, "inconsistent bwt_size"); xassert(k == fmt->bwt_size, "inconsistent bwt_size");
// update fmt // update fmt
fmt->bwt = buf; fmt->bwt = buf;
@ -386,7 +380,7 @@ void fmt_e2_occ(const FMTIndex *fmt, bwtint_t k, int b1, int b2, bwtint_t cnt[4]
cnt[3] = q[b2]; // b2的occ cnt[3] = q[b2]; // b2的occ
p += 20; p += 20;
#ifdef FMT_MID_INTERVAL #ifdef FMT_MID_INTERVAL // 加入了middle checkpoint
// 使用mid interval信息 // 使用mid interval信息
int mk = k % FMT_OCC_INTERVAL; int mk = k % FMT_OCC_INTERVAL;
int n_mintv = mk >> FMT_MID_INTV_SHIFT; int n_mintv = mk >> FMT_MID_INTV_SHIFT;
@ -405,29 +399,34 @@ void fmt_e2_occ(const FMTIndex *fmt, bwtint_t k, int b1, int b2, bwtint_t cnt[4]
x = 0; x = 0;
p += 4; p += 4;
} }
#if FMT_MID_INTERVAL == 16 #if FMT_MID_INTERVAL == 16 // middle check point interval等于16时候，只需要判断一下是不是要计算两个uint32表示的碱基序列
if ((mk & FMT_MID_INTV_MASK) >> 3) if ((mk & FMT_MID_INTV_MASK) >> 3)
{ {
x += __fmt_occ_e2_aux2(fmt, ti, *p); x += __fmt_occ_e2_aux2(fmt, ti, *p);
++p; ++p;
} }
#elif FMT_MID_INTERVAL > 16 // 该地址是bwt和pre_bwt字符串数据的首地址 #elif FMT_MID_INTERVAL > 16 // 该地址是bwt和pre_bwt字符串数据的首地址
uint32_t *end = p + ((k >> 3) - ((k & ~FMT_OCC_INTV_MASK) >> 3)); uint32_t *end = p + ((k >> 3) - ((k & ~FMT_MID_INTV_MASK) >> 3));
for (; p < end; ++p) for (; p < end; ++p)
{ {
x += __fmt_occ_e2_aux2(fmt, ti, *p); x += __fmt_occ_e2_aux2(fmt, ti, *p);
} }
tmp = *p & ~((1U << ((~k & 7) << 2)) - 1);
x += __fmt_occ_e2_aux2(fmt, ti, tmp);
#endif #endif
#else #else // 没有加入middle check point interval
#if FMT_OCC_INTERVAL > 16
uint32_t *end = p + ((k >> 3) - ((k & ~FMT_OCC_INTV_MASK) >> 3)); uint32_t *end = p + ((k >> 3) - ((k & ~FMT_OCC_INTV_MASK) >> 3));
// p = end - (end - p) / 4;
for (; p < end; ++p) for (; p < end; ++p)
{ {
x += __fmt_occ_e2_aux2(fmt, ti, *p); x += __fmt_occ_e2_aux2(fmt, ti, *p);
} }
tmp = *p & ~((1U << ((~k & 7) << 2)) - 1); #else // FMT_OCC_INTERVAL等于16的时候，只需要判断一次就可以
x += __fmt_occ_e2_aux2(fmt, ti, tmp); if ((k & FMT_OCC_INTV_MASK) >> 3)
{
x += __fmt_occ_e2_aux2(fmt, ti, *p);
++p;
}
#endif
#endif #endif
tmp = *p & ~((1U << ((~k & 7) << 2)) - 1); tmp = *p & ~((1U << ((~k & 7) << 2)) - 1);
x += __fmt_occ_e2_aux2(fmt, ti, tmp); x += __fmt_occ_e2_aux2(fmt, ti, tmp);
@ -452,119 +451,13 @@ void fmt_e2_occ(const FMTIndex *fmt, bwtint_t k, int b1, int b2, bwtint_t cnt[4]
cnt[3] += x >> 24 & 0xff; cnt[3] += x >> 24 & 0xff;
} }
void calc_intv_occ(const FMTIndex *fmt, bwtint_t k, bwtint_t str_line, bwtint_t cp_line,
uint32_t *p, int b1, int b2, bwtint_t cnt[4])
{
uint32_t x = 0;
uint32_t *q, tmp;
int i, ti = b1 << 2 | b2;
#ifdef FMT_MID_INTERVAL
// 使用mid interval信息
int mk = k % FMT_OCC_INTERVAL;
int n_mintv = mk >> FMT_MID_INTV_SHIFT;
if (n_mintv > 0) // 至少超过了第一个mid interval
{
p += n_mintv * (4 + (FMT_MID_INTERVAL >> 3)) - 4; // 对应的mid interval check point的首地址，即A C G T的局部累积量
q = p + b1;
for (i = b1 + 1; i < 4; ++i)
x += p[i]; // 大于b1的碱基的occ之和
cnt[0] += __fmt_mid_sum(x);
x = *q;
cnt[1] += __fmt_mid_sum(x); // b1的occ
for (i = 3; i > b2; --i)
cnt[2] += x >> (i << 3) & 0xff; // 大于b2的occ之和
cnt[3] += x >> (b2 << 3) & 0xff; // b2的occ
x = 0;
p += 4;
}
#if FMT_MID_INTERVAL == 16
if ((mk & FMT_MID_INTV_MASK) >> 3)
{
x += __fmt_occ_e2_aux2(fmt, ti, *p);
++p;
}
#elif FMT_MID_INTERVAL > 16 // 该地址是bwt和pre_bwt字符串数据的首地址
uint32_t *end = p + ((k >> 3) - ((k & ~FMT_OCC_INTV_MASK) >> 3));
for (; p < end; ++p)
{
x += __fmt_occ_e2_aux2(fmt, ti, *p);
}
tmp = *p & ~((1U << ((~k & 7) << 2)) - 1);
x += __fmt_occ_e2_aux2(fmt, ti, tmp);
#endif
#else
uint32_t *end = p + ((k >> 3) - ((k & ~FMT_OCC_INTV_MASK) >> 3));
for (; p < end; ++p)
{
x += __fmt_occ_e2_aux2(fmt, ti, *p);
}
tmp = *p & ~((1U << ((~k & 7) << 2)) - 1);
x += __fmt_occ_e2_aux2(fmt, ti, tmp);
#endif
tmp = *p & ~((1U << ((~k & 7) << 2)) - 1);
x += __fmt_occ_e2_aux2(fmt, ti, tmp);
if (b1 == 0)
{
x -= (~k & 7) << 8;
if (b2 == 0)
x -= (~k & 7) << 24;
}
// 如果跨过了second_primary,那么可能需要减掉一次累积值
if (b1 == fmt->first_base && cp_line < fmt->sec_primary && str_line >= fmt->sec_primary)
{
if (b2 < fmt->last_base)
cnt[2] -= 1;
else if (b2 == fmt->last_base)
cnt[3] -= 1;
}
cnt[0] += x & 0xff;
cnt[1] += x >> 8 & 0xff;
cnt[2] += x >> 16 & 0xff;
cnt[3] += x >> 24 & 0xff;
}
void fmt_e2_2occ(const FMTIndex *fmt, bwtint_t k, bwtint_t l, int b1, int b2, bwtint_t tk[4], bwtint_t tl[4])
{
bwtint_t _k, _l;
_k = k - (k >= fmt->primary);
_l = l - (l >= fmt->primary);
if (_l >> FMT_OCC_INTV_SHIFT != _k >> FMT_OCC_INTV_SHIFT || k == (bwtint_t)(-1) || l == (bwtint_t)(-1))
{
fmt_e2_occ(fmt, k, b1, b2, tk);
fmt_e2_occ(fmt, l, b1, b2, tl);
}
else
{
uint32_t *p, *q;
bwtint_t cp_line = _k >> FMT_OCC_INTV_SHIFT << FMT_OCC_INTV_SHIFT;
int i;
tk[0] = 0;
tk[2] = 0;
p = fmt_occ_intv(fmt, _k);
for (i = b1 + 1; i < 4; ++i)
tk[0] += p[i]; // 大于b1的碱基的occ之和
tk[1] = p[b1]; // b1的occ
q = p + 4 + b1 * 4;
for (i = b2 + 1; i < 4; ++i)
tk[2] += q[i]; // 大于b2的occ之和
tk[3] = q[b2]; // b2的occ
p += 20;
memcpy(tl, tk, 32);
calc_intv_occ(fmt, _k, k, cp_line, p, b1, b2, tk);
calc_intv_occ(fmt, _l, l, cp_line, p, b1, b2, tl);
}
}
// 扩展两个碱基 // 扩展两个碱基
void fmt_extend2(const FMTIndex *fmt, bwtintv_t *ik, bwtintv_t *ok, int is_back, int b1, int b2) void fmt_extend2(const FMTIndex *fmt, bwtintv_t *ik, bwtintv_t *ok, int is_back, int b1, int b2)
{ {
bwtint_t tk[4], tl[4], first_pos; bwtint_t tk[4], tl[4], first_pos;
// tk表示在k行之前所有各个碱基累积出现次数，tl表示在l行之前的累积 // tk表示在k行之前所有各个碱基累积出现次数，tl表示在l行之前的累积
//fmt_e2_occ(fmt, ik->x[!is_back] - 1, b1, b2, tk); fmt_e2_occ(fmt, ik->x[!is_back] - 1, b1, b2, tk);
//fmt_e2_occ(fmt, ik->x[!is_back] - 1 + ik->x[2], b1, b2, tl); fmt_e2_occ(fmt, ik->x[!is_back] - 1 + ik->x[2], b1, b2, tl);
fmt_e2_2occ(fmt, ik->x[!is_back] - 1, ik->x[!is_back] - 1 + ik->x[2], b1, b2, tk, tl);
//fmt_e2_occ_2way(fmt, ik->x[!is_back] - 1, b1, b2, tk); //fmt_e2_occ_2way(fmt, ik->x[!is_back] - 1, b1, b2, tk);
//fmt_e2_occ_2way(fmt, ik->x[!is_back] - 1 + ik->x[2], b1, b2, tl); //fmt_e2_occ_2way(fmt, ik->x[!is_back] - 1 + ik->x[2], b1, b2, tl);
@ -587,9 +480,8 @@ void fmt_extend1(const FMTIndex *fmt, bwtintv_t *ik, bwtintv_t *ok, int is_back,
bwtint_t tk[4], tl[4]; bwtint_t tk[4], tl[4];
int b2 = 3; int b2 = 3;
// tk表示在k行之前所有各个碱基累积出现次数，tl表示在l行之前的累积 // tk表示在k行之前所有各个碱基累积出现次数，tl表示在l行之前的累积
//fmt_e2_occ(fmt, ik->x[!is_back] - 1, b1, b2, tk); fmt_e2_occ(fmt, ik->x[!is_back] - 1, b1, b2, tk);
//fmt_e2_occ(fmt, ik->x[!is_back] - 1 + ik->x[2], b1, b2, tl); fmt_e2_occ(fmt, ik->x[!is_back] - 1 + ik->x[2], b1, b2, tl);
fmt_e2_2occ(fmt, ik->x[!is_back] - 1, ik->x[!is_back] - 1 + ik->x[2], b1, b2, tk, tl);
//fmt_e2_occ_2way(fmt, ik->x[!is_back] - 1, b1, b2, tk); //fmt_e2_occ_2way(fmt, ik->x[!is_back] - 1, b1, b2, tk);
//fmt_e2_occ_2way(fmt, ik->x[!is_back] - 1 + ik->x[2], b1, b2, tl); //fmt_e2_occ_2way(fmt, ik->x[!is_back] - 1 + ik->x[2], b1, b2, tl);
// 这里是反向扩展 // 这里是反向扩展
@ -655,9 +547,9 @@ uint32_t str2bit(string &str)
int main_fmtidx(int argc, char **argv) int main_fmtidx(int argc, char **argv)
{ {
// string seq("ACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTA"); // string seq("ACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTA");
string seq("ACCCT"); //string seq("ACCCT");
string rseq = calc_reverse_seq(seq); //string rseq = calc_reverse_seq(seq);
seq = seq + rseq; //seq = seq + rseq;
//create_bwt_mtx(seq); //create_bwt_mtx(seq);
//cout << seq << endl; //cout << seq << endl;
@ -682,92 +574,93 @@ int main_fmtidx(int argc, char **argv)
// } // }
// return 0; // return 0;
string prefix = argv[1];
string bwt_str = prefix + ".bwt.str";
string bwt_idx, fmt_idx;
bwt_t *bwt;
FMTIndex *fmt;
ostringstream oss_bwt, oss_fmt;
oss_bwt << '.' << OCC_INTERVAL;
oss_fmt << '.' << FMT_OCC_INTERVAL;
#ifdef FMT_MID_INTERVAL
oss_fmt << '.' << FMT_MID_INTERVAL;
#endif
// string bwt_str = string(argv[1]) + ".bwt.str"; bwt_idx = prefix + oss_bwt.str() + ".bwt";
string bwt_idx = string(argv[1]) + ".128.bwt"; fmt_idx = prefix + oss_fmt.str() + ".fmt";
// string bwt_idx = string(argv[1]) + ".64.bwt";
// string bwt_idx = string(argv[1]) + ".16.bwt";
// string bwt_idx = string(argv[1]) + ".bwt";
//bwt_t *bwt = restore_bwt(bwt_str.c_str());
//create_interval_occ_bwt(bwt);
//dump_bwt(bwt_idx.c_str(), bwt);
//string fmt_idx = string(argv[1]) + ".fmt"; cout << bwt_idx << endl;
//string fmt_idx = string(argv[1]) + ".256.8.fmt"; cout << fmt_idx << endl;
string fmt_idx = string(argv[1]) + ".256.fmt";
//string fmt_idx = string(argv[1]) + ".128.fmt";
// string fmt_idx = string(argv[1]) + ".64.fmt";
// string fmt_idx = string(argv[1]) + ".32.fmt";
t11 = realtime(); // 生成或读取bwt索引文件
bwt_t *bwt = restore_bwt(bwt_idx.c_str()); double time_read_bwt = realtime();
t11 = realtime() - t11; #if 0
cout << "[time read bwt:] " << t11 << "s" << endl; bwt = restore_bwt(bwt_str.c_str());
t10 = realtime(); create_interval_occ_bwt(bwt);
FMTIndex *fmt = restore_fmt(fmt_idx.c_str()); dump_bwt(bwt_idx.c_str(), bwt);
//FMTIndex *fmt = create_fmt_from_bwt(bwt); #else
//dump_fmt(fmt_idx.c_str(), fmt); bwt = restore_bwt(bwt_idx.c_str());
t10 = realtime() - t10; #endif
cout << "[time gen fmt:] " << t10 << "s" << endl; time_read_bwt = realtime() - time_read_bwt;
cout << "[time gen/read bwt:] " << time_read_bwt << "s" << endl;
vector<string> seed_arr(10000000); // 生成或读取fmt索引文件
seed_arr[0] = "GCGATACTAAGA"; double time_read_fmt = realtime();
seed_arr[0] = "GAGAGCTGTTCCCGTGTTTTCCATGGTTT"; #if 0
fmt = create_fmt_from_bwt(bwt);
dump_fmt(fmt_idx.c_str(), fmt);
#else
fmt = restore_fmt(fmt_idx.c_str());
#endif
time_read_fmt = realtime() - time_read_fmt;
cout << "[time gen/read fmt:] " << time_read_fmt << "s" << endl;
// 生成随机序列进行测试
int seq_size = 10000000;
// int seq_size = 1;
int seq_len = 19;
vector<string> seq_arr(seq_size);
seq_arr[0] = "GCGATACTAAGA";
// seq_arr[0] = "GAGAGCTGTTCCCGTGTTTTCCATGGTTT";
#if 1
srand(time(NULL)); srand(time(NULL));
double time_gen_seq = realtime();
for (int i = 0; i < (int)seq_arr.size(); ++i)
seq_arr[i] = generate_rand_seq(seq_len);
time_gen_seq = realtime() - time_gen_seq;
cout << "[time gen seq:] " << time_gen_seq << "s" << endl;
#endif
t1 = realtime(); // 对比bwt和fmt搜索的时间
for (int i = 0; i < (int)seed_arr.size(); ++i) #if 1
seed_arr[i] = generate_rand_seq(29); double time_bwt_search = realtime();
t1 = realtime() - t1; for (int i = 0; i < (int)seq_arr.size(); ++i)
cout << "[time gen seed:] " << t1 << "s" << endl; bwt_search(bwt, seq_arr[i]);
time_bwt_search = realtime() - time_bwt_search;
cout << "[time bwt search:] " << time_bwt_search << "s" << endl;
t2 = realtime(); double time_fmt_search = realtime();
for (int i = 0; i < (int)seed_arr.size(); ++i) for (int i = 0; i < (int)seq_arr.size(); ++i)
bwt_search(bwt, seed_arr[i]); fmt_search(fmt, seq_arr[i]);
t2 = realtime() - t2; time_fmt_search = realtime() - time_fmt_search;
cout << "[time bwt search:] " << t2 << "s" << endl; cout << "[time fmt search:] " << time_fmt_search << "s" << endl;
t3 = realtime(); #endif
for (int i = 0; i < (int)seed_arr.size(); ++i)
fmt_search(fmt, seed_arr[i]);
t3 = realtime() - t3;
cout << "[time fmt search:] " << t3 << "s" << endl;
//return 0; // 对比bwt和fmt搜索的结果
#if 1
t4 = realtime(); double time_cmp = realtime();
for (int i = 0; i < (int)seed_arr.size(); ++i) for (int i = 0; i < (int)seq_arr.size(); ++i)
{ {
bwtintv_t p1 = bwtintv_t p1 = bwt_search2(bwt, seq_arr[i]);
bwt_search2(bwt, seed_arr[i]); bwtintv_t p2 = fmt_search(fmt, seq_arr[i]);
bwtintv_t p2 = fmt_search(fmt, seed_arr[i]);
if (p1.x[0] != p2.x[0] || p1.x[1] != p2.x[1] || p1.x[2] != p2.x[2]) if (p1.x[0] != p2.x[0] || p1.x[1] != p2.x[1] || p1.x[2] != p2.x[2])
cout << seed_arr[i] << endl cout << seq_arr[i] << endl
<< p1.x[0] << ' ' << p1.x[1] << ' ' << p1.x[2] << endl << p1.x[0] << ' ' << p1.x[1] << ' ' << p1.x[2] << endl
<< p2.x[0] << ' ' << p2.x[1] << ' ' << p2.x[2] << endl; << p2.x[0] << ' ' << p2.x[1] << ' ' << p2.x[2] << endl;
} }
t4 = realtime() - t4; time_cmp = realtime() - time_cmp;
cout << "[time bwt search 2:] " << t4 << "s" << endl; cout << "[time compare bwt & fmt:] " << time_cmp << "s" << endl;
#endif
//cout << "bwt occ: " << t5 << "s; " << t7 << '\t' << t10 << endl;
//cout << "fmt occ: " << t6 << "s; " << t11 << '\t' << t8 << '\t' << t9 << endl;
// bwt_search(bwt, s);
// bwt_search2(bwt, s);
// for (int i = 0; i < 120; ++i)
// {
// cout << i << '\t' << bwt_B0(bwt, i) << endl;
// }
// TGGGAT
// FMTIndex *fmt = create_fmt_from_bwt(bwt);
// dump_fmt("ref.fmt", fmt);
// FMTIndex *fmt = restore_fmt("tiny.fmt");
//fmt_search(fmt, s);
// cout << bwt->bwt_size << endl;
// cout << bwt->seq_len << endl;
//cout << "sec_: " << fmt->sec_bcp << '\t' << fmt->sec_primary << endl;
//uint8_t b8 = 2 << 4 | 2;
//cout << "AGAG: " << fmt->cnt_table[2][b8] << endl;
//cout << (((b8 >> 6) == 0 && (b8 >> 4 & 3) == 2) + ((b8 >> 2 & 3) == 0 && (b8 & 3) == 2)) << endl;
return 0; return 0;
} }

18
fmt_index.h
View File

@ -5,18 +5,26 @@
#define FMT_OCC_INTERVAL (1LL << FMT_OCC_INTV_SHIFT) #define FMT_OCC_INTERVAL (1LL << FMT_OCC_INTV_SHIFT)
#define FMT_OCC_INTV_MASK (FMT_OCC_INTERVAL - 1) #define FMT_OCC_INTV_MASK (FMT_OCC_INTERVAL - 1)
#define FMT_MID_INTV_SHIFT 4 #define FMT_MID_INTV_SHIFT 7
#define FMT_MID_INTERVAL (1LL << FMT_MID_INTV_SHIFT) #define FMT_MID_INTERVAL (1LL << FMT_MID_INTV_SHIFT)
#define FMT_MID_INTV_MASK (FMT_MID_INTERVAL - 1) #define FMT_MID_INTV_MASK (FMT_MID_INTERVAL - 1)
// #undef FMT_MID_INTERVAL
// 获取碱基c（待查找序列的首个碱基）和对应的互补碱基对应的行，以及间隔 // 获取碱基c（待查找序列的首个碱基）和对应的互补碱基对应的行，以及间隔
#define fmt_set_intv(fmt, c, ik) ((ik).x[0] = (fmt)->L2[(int)(c)] + 1, (ik).x[2] = (fmt)->L2[(int)(c) + 1] - (fmt)->L2[(int)(c)], (ik).x[1] = (fmt)->L2[3 - (c)] + 1, (ik).info = 0) #define fmt_set_intv(fmt, c, ik) ((ik).x[0] = (fmt)->L2[(int)(c)] + 1, (ik).x[2] = (fmt)->L2[(int)(c) + 1] - (fmt)->L2[(int)(c)], (ik).x[1] = (fmt)->L2[3 - (c)] + 1, (ik).info = 0)
// k行（bwt str行（不包含$））对应的check point occ数据起始地址（小于k且是OCC_INTERVAL的整数倍） // k行（bwt str行（不包含$））对应的check point occ数据起始地址（小于k且是OCC_INTERVAL的整数倍）
#if FMT_MID_INTERVAL == 16
// 包含mid interval #if FMT_MID_INTERVAL == 8
#define fmt_occ_intv(b, k) ((b)->bwt + (k) / FMT_OCC_INTERVAL * (FMT_OCC_INTERVAL / 8 + 80))
#elif FMT_MID_INTERVAL == 8
#define fmt_occ_intv(b, k) ((b)->bwt + (k) / FMT_OCC_INTERVAL * (FMT_OCC_INTERVAL / 8 + 144)) #define fmt_occ_intv(b, k) ((b)->bwt + (k) / FMT_OCC_INTERVAL * (FMT_OCC_INTERVAL / 8 + 144))
#elif FMT_MID_INTERVAL == 16
#define fmt_occ_intv(b, k) ((b)->bwt + (k) / FMT_OCC_INTERVAL * (FMT_OCC_INTERVAL / 8 + 80))
#elif FMT_MID_INTERVAL == 32
#define fmt_occ_intv(b, k) ((b)->bwt + (k) / FMT_OCC_INTERVAL * (FMT_OCC_INTERVAL / 8 + 48))
#elif FMT_MID_INTERVAL == 64
#define fmt_occ_intv(b, k) ((b)->bwt + (k) / FMT_OCC_INTERVAL * (FMT_OCC_INTERVAL / 8 + 32))
#elif FMT_MID_INTERVAL == 128
#define fmt_occ_intv(b, k) ((b)->bwt + (k) / FMT_OCC_INTERVAL * (FMT_OCC_INTERVAL / 8 + 24))
#else #else
#define fmt_occ_intv(b, k) ((b)->bwt + (k) / FMT_OCC_INTERVAL * (FMT_OCC_INTERVAL / 8 + 20)) #define fmt_occ_intv(b, k) ((b)->bwt + (k) / FMT_OCC_INTERVAL * (FMT_OCC_INTERVAL / 8 + 20))
#endif #endif