edit distance based error correction

This commit is contained in:
Haoyu Cheng
2019-07-07 03:44:59 -04:00
parent b553f3ac5e
commit 87873d6323
35 changed files with 3910 additions and 6 deletions

View File

@@ -237,6 +237,111 @@ void init_UC_Read(UC_Read* r)
}
void recover_UC_Read_sub_region(char* r, long long start_pos, long long length, uint8_t strand, All_reads* R_INF, long long ID)
{
long long readLen = Get_READ_LENGTH((*R_INF), ID);
uint8_t* src = Get_READ((*R_INF), ID);
long long i;
long long copyLen;
long long end_pos = start_pos + length - 1;
if (strand == 0)
{
i = start_pos;
copyLen = 0;
long long initLen = start_pos % 4;
if (initLen != 0)
{
memcpy(r, bit_t_seq_table[src[i>>2]] + initLen, 4 - initLen);
copyLen = copyLen + 4 - initLen;
i = i + copyLen;
}
while (copyLen < length)
{
memcpy(r+copyLen, bit_t_seq_table[src[i>>2]], 4);
copyLen = copyLen + 4;
i = i + 4;
}
if (R_INF->N_site[ID])
{
for (i = 1; i <= R_INF->N_site[ID][0]; i++)
{
if (R_INF->N_site[ID][i] >= start_pos && R_INF->N_site[ID][i] <= end_pos)
{
r[R_INF->N_site[ID][i] - start_pos] = 'N';
}
else if(R_INF->N_site[ID][i] > end_pos)
{
break;
}
}
}
}
else
{
start_pos = readLen - start_pos - 1;
end_pos = readLen - end_pos - 1;
///start_pos > end_pos
i = start_pos;
copyLen = 0;
long long initLen = (start_pos + 1) % 4;
if (initLen != 0)
{
memcpy(r, bit_t_seq_table_rc[src[i>>2]] + 4 - initLen, initLen);
copyLen = copyLen + initLen;
i = i - initLen;
}
while (copyLen < length)
{
memcpy(r+copyLen, bit_t_seq_table_rc[src[i>>2]], 4);
copyLen = copyLen + 4;
i = i - 4;
}
if (R_INF->N_site[ID])
{
long long offset = readLen - start_pos - 1;
for (i = 1; i <= R_INF->N_site[ID][0]; i++)
{
if (R_INF->N_site[ID][i] >= end_pos && R_INF->N_site[ID][i] <= start_pos)
{
r[readLen - R_INF->N_site[ID][i] - 1 - offset] = 'N';
}
else if(R_INF->N_site[ID][i] > start_pos)
{
break;
}
}
}
}
}
void recover_UC_Read(UC_Read* r, All_reads* R_INF, uint64_t ID)
{
r->length = Get_READ_LENGTH((*R_INF), ID);