END_TEST
START_TEST (ut_graph_load_dictionary_prefix_search)
{
init_node(graph);
FILE *fp = fopen("/usr/share/dict/words", "r");
wchar_t word[1000];
while (fgetws(word, 80, fp))
{
word[wcslen(word)-1] = L'\0';
printf("%ls\n", word);
insert_word(graph, word);
}
fclose(fp);
PrefixResult *result = search_prefix(graph, L"app");
while(result != NULL)
{
printf("%ls\n", result->word);
result = result->next;
}
clear_node(graph);
}
void suffix_mng_search_read_cals_by_region(fastq_read_t *read, int num_seeds,
sa_index3_t *sa_index,
int strand, int chromosome,
size_t start, size_t end,
array_list_t *cal_list,
suffix_mng_t *suffix_mng) {
int chrom, num_prefixes, num_suffixes, suffix_len = 0;
size_t low, high, r_start_suf, r_end_suf, g_start_suf, g_end_suf;
int read_pos, read_inc = read->length / num_seeds;
if (read_inc < sa_index->k_value / 2) {
read_inc = sa_index->k_value / 2;
}
// Fill in the CAL manager structure
int read_end_pos = read->length - sa_index->k_value;
int extra_seed = (read->length - sa_index->k_value) % read_inc;
// first step, searching mappings in both strands
// distance between seeds >= prefix value (sa_index->k_value)
char *r_seq = strand == 0 ? read->sequence : read->revcomp;
for (read_pos = 0; read_pos < read_end_pos; read_pos += read_inc) {
num_prefixes = search_prefix(&r_seq[read_pos], &low, &high, sa_index, 0);
num_suffixes = num_prefixes;
suffix_len = num_suffixes > 0 ? sa_index->k_value : 0;
if (num_suffixes > 0) {
for (size_t suff = low; suff <= high; suff++) {
chrom = sa_index->CHROM[suff];
if (chrom == chromosome) {
// extend suffix to right side
r_start_suf = read_pos;
r_end_suf = r_start_suf + suffix_len - 1;
g_start_suf = sa_index->SA[suff] - sa_index->genome->chrom_offsets[chrom];
g_end_suf = g_start_suf + suffix_len - 1;
if (start <= g_start_suf && end >= g_end_suf) {
suffix_mng_update(chrom, r_start_suf, r_end_suf, g_start_suf, g_end_suf, suffix_mng);
}
}
}
}
} // end of for read_pos
if (suffix_len != read->length && extra_seed) {
read_pos = read->length - sa_index->k_value;
num_prefixes = search_prefix(&r_seq[read_pos], &low, &high, sa_index, 0);
num_suffixes = num_prefixes;
suffix_len = num_suffixes > 0 ? sa_index->k_value : 0;
if (num_suffixes > 0) {
for (size_t suff = low; suff <= high; suff++) {
chrom = sa_index->CHROM[suff];
if (chrom == chromosome) {
// extend suffix to right side
r_start_suf = read_pos;
r_end_suf = r_start_suf + suffix_len - 1;
g_start_suf = sa_index->SA[suff] - sa_index->genome->chrom_offsets[chrom];
g_end_suf = g_start_suf + suffix_len - 1;
if (start <= g_start_suf && end >= g_end_suf) {
suffix_mng_update(chrom, r_start_suf, r_end_suf, g_start_suf, g_end_suf, suffix_mng);
}
}
}
}
}
// using suffix manager instead of the previous cal manager
suffix_mng_create_cals(read, read->length / 3, strand,
sa_index, cal_list, suffix_mng);
}