void align(char* seq_a, char* seq_b)
{
// Variables to store alignment result
sw_aligner_t *sw = smith_waterman_new();
alignment_t *result = alignment_create(256);
// Decide on scoring
int match = 1;
int mismatch = -2;
int gap_open = -4;
int gap_extend = -1;
// Don't penalise gaps at the start
// ACGATTT
// ----TTT would score +3 (when match=+1)
char no_start_gap_penalty = 1;
// ..or gaps at the end e.g.
// ACGATTT
// ACGA--- would score +4 (when match=+1)
char no_end_gap_penalty = 1;
char no_gaps_in_a = 0, no_gaps_in_b = 0;
char no_mismatches = 0;
// Compare character case-sensitively (usually set to 0 for DNA etc)
char case_sensitive = 0;
scoring_t scoring;
scoring_init(&scoring, match, mismatch, gap_open, gap_extend,
no_start_gap_penalty, no_end_gap_penalty,
no_gaps_in_a, no_gaps_in_b, no_mismatches, case_sensitive);
// Add some special cases
// x -> y means x in seq1 changing to y in seq2
scoring_add_mutation(&scoring, 'a', 'c', -2); // a -> c give substitution score -2
scoring_add_mutation(&scoring, 'c', 'a', -1); // c -> a give substitution score -1
// We could also prohibit the aligning of characters not given as special cases
// scoring.use_match_mismatch = 0;
smith_waterman_align(seq_a, seq_b, &scoring, sw);
while(smith_waterman_fetch(sw, result))
{
printf("seqA: %s [start:%zu]\n", result->result_a, result->pos_a);
printf("seqB: %s [start:%zu]\n", result->result_b, result->pos_b);
printf("alignment score: %i\n\n", result->score);
}
// Free memory for storing alignment results
smith_waterman_free(sw);
alignment_free(result);
}
int main(int argc, char* argv[])
{
#ifdef SEQ_ALIGN_VERBOSE
printf("VERBOSE: on\n");
#endif
sw_set_default_scoring();
cmd = cmdline_new(argc, argv, &scoring, SEQ_ALIGN_SW_CMD);
// Align!
sw = smith_waterman_new();
result = alignment_create(256);
if(cmd->seq1 != NULL)
{
// Align seq1 and seq2
align(cmd->seq1, cmd->seq2, NULL, NULL);
}
// Align from files
size_t i, num_of_file_pairs = cmdline_get_num_of_file_pairs(cmd);
for(i = 0; i < num_of_file_pairs; i++)
{
const char *file1 = cmdline_get_file1(cmd, i);
const char *file2 = cmdline_get_file2(cmd, i);
if(file1 != NULL && *file1 == '\0' && file2 == NULL) {
wait_on_keystroke = 1;
file1 = "-";
}
align_from_file(file1, file2, &align_pair_from_file, !cmd->interactive);
}
// Free memory for storing alignment results
smith_waterman_free(sw);
alignment_free(result);
cmdline_free(cmd);
return EXIT_SUCCESS;
}
int main(int argc, char **argv)
{
if(argc != 2) print_usage(argv);
char *seq = argv[1];
size_t seqlen = strlen(seq);
// Go
int match = 1, mismatch = -1, gap_open = -4, gap_extend = -1;
bool no_start_gap_penalty = false, no_end_gap_penalty = false;
bool no_gaps_in_a = true, no_gaps_in_b = true;
bool no_mismatches = true, case_sensitive = true;
scoring_t scoring;
scoring_init(&scoring, match, mismatch, gap_open, gap_extend,
no_start_gap_penalty, no_end_gap_penalty,
no_gaps_in_a, no_gaps_in_b,
no_mismatches, case_sensitive);
// Alignment results stored here
sw_aligner_t *sw = smith_waterman_new();
alignment_t *aln = alignment_create(seqlen+1);
smith_waterman_align(seq, seq, &scoring, sw);
// Loop over results
while(smith_waterman_fetch(sw, aln))
{
if(aln->pos_a < aln->pos_b) {
fputs(aln->result_a, stdout);
printf(" [%zu,%zu]\n", aln->pos_a, aln->pos_b);
}
}
smith_waterman_free(sw);
alignment_free(aln);
return EXIT_SUCCESS;
}
void sw_test_no_gaps_smith_waterman()
{
sw_aligner_t *sw = smith_waterman_new();
alignment_t *result = alignment_create(256);
const char* seq_a = "gacag";
const char* seq_b = "tgaagt";
int match = 1;
int mismatch = -2;
int gap_open = -4;
int gap_extend = -1;
bool no_start_gap_penalty = false, no_end_gap_penalty = false;
bool no_gaps_in_a = true, no_gaps_in_b = true;
bool no_mismatches = false, case_sensitive = true;
scoring_t scoring;
scoring_init(&scoring, match, mismatch, gap_open, gap_extend,
no_start_gap_penalty, no_end_gap_penalty,
no_gaps_in_a, no_gaps_in_b,
no_mismatches, case_sensitive);
smith_waterman_align(seq_a, seq_b, &scoring, sw);
smith_waterman_fetch(sw, result);
ASSERT(strcmp(result->result_a, "ga") == 0 &&
strcmp(result->result_b, "ga") == 0);
smith_waterman_fetch(sw, result);
ASSERT(strcmp(result->result_a, "ag") == 0 &&
strcmp(result->result_b, "ag") == 0);
alignment_free(result);
smith_waterman_free(sw);
}
