END_TEST
START_TEST (number_of_snps_detected_small)
{
detect_snps("../tests/data/small_alignment.aln");
fail_unless( get_number_of_snps() == 1);
}
END_TEST
START_TEST (number_of_snps_detected)
{
detect_snps("../tests/data/alignment_file_multiple_lines_per_sequence.aln");
fail_unless( get_number_of_snps() == 5);
}
END_TEST
START_TEST (valid_number_of_sequences_in_file_with_multiple_lines_per_sequence)
{
detect_snps("../tests/data/alignment_file_multiple_lines_per_sequence.aln");
fail_unless( get_number_of_samples() == 109 );
}
END_TEST
START_TEST (valid_genome_length_with_multiple_lines_per_sequence)
{
detect_snps("../tests/data/alignment_file_multiple_lines_per_sequence.aln");
fail_unless( get_length_of_genome() == 2000 );
}
END_TEST
START_TEST (sample_names_from_alignment_file)
{
detect_snps("../tests/data/small_alignment.aln");
char ** current_sequence_names = get_sequence_names();
fail_unless(strcmp(current_sequence_names[0],"reference_sequence") == 0);
fail_unless(strcmp(current_sequence_names[1],"comparison_sequence") == 0);
fail_unless(strcmp(current_sequence_names[2],"another_comparison_sequence") == 0);
}
static int generate_snp_sites_generic(char filename[],
int output_multi_fasta_file,
int output_vcf_file,
int output_phylip_file,
char output_filename[],
int output_reference, int pure_mode, int output_monomorphic)
{
int i;
detect_snps(filename, pure_mode, output_monomorphic);
bases_for_snps = calloc(get_number_of_snps()+1, sizeof(char*));
for(i = 0; i < get_number_of_snps(); i++)
{
bases_for_snps[i] = calloc(get_number_of_samples()+1, sizeof(char));
}
get_bases_for_each_snp(filename, bases_for_snps);
char output_filename_base[FILENAME_MAX];
char filename_without_directory[FILENAME_MAX];
strip_directory_from_filename(filename, filename_without_directory);
strncpy(output_filename_base, filename_without_directory, FILENAME_MAX);
if(output_filename != NULL && *output_filename != '\0')
{
strncpy(output_filename_base, output_filename, FILENAME_MAX);
}
if(output_vcf_file)
{
char vcf_output_filename[FILENAME_MAX];
strncpy(vcf_output_filename, output_filename_base, FILENAME_MAX);
if((output_vcf_file + output_phylip_file + output_multi_fasta_file) > 1 || (output_filename == NULL || *output_filename == '\0') )
{
strcat(vcf_output_filename, ".vcf");
}
create_vcf_file(vcf_output_filename, get_snp_locations(), get_number_of_snps(), bases_for_snps, get_sequence_names(), get_number_of_samples(), get_length_of_genome(), get_pseudo_reference_sequence());
}
if(output_phylip_file)
{
char phylip_output_filename[FILENAME_MAX];
strncpy(phylip_output_filename, output_filename_base, FILENAME_MAX);
if((output_vcf_file + output_phylip_file + output_multi_fasta_file) > 1 || (output_filename == NULL || *output_filename == '\0') )
{
strcat(phylip_output_filename, ".phylip");
}
create_phylib_of_snp_sites(phylip_output_filename, get_number_of_snps(), bases_for_snps, get_sequence_names(), get_number_of_samples(), output_reference, get_pseudo_reference_sequence(),get_snp_locations());
}
if((output_multi_fasta_file) || (output_vcf_file ==0 && output_phylip_file == 0 && output_multi_fasta_file == 0))
{
char multi_fasta_output_filename[FILENAME_MAX];
strncpy(multi_fasta_output_filename, output_filename_base, FILENAME_MAX);
if((output_vcf_file + output_phylip_file + output_multi_fasta_file) > 1 || (output_filename == NULL || *output_filename == '\0') )
{
strcat(multi_fasta_output_filename, ".snp_sites.aln");
}
create_fasta_of_snp_sites(multi_fasta_output_filename, get_number_of_snps(), bases_for_snps, get_sequence_names(), get_number_of_samples(), output_reference, get_pseudo_reference_sequence(),get_snp_locations());
}
// free memory
free(get_snp_locations());
for(i = 0; i < get_number_of_samples(); i++)
{
// free(get_sequence_names().[i]);
}
for(i = 0; i < get_number_of_snps(); i++)
{
free(bases_for_snps[i]);
}
free(get_pseudo_reference_sequence());
return 1;
}
int generate_snp_sites(char filename[],int output_multi_fasta_file, int output_vcf_file, int output_phylip_file, char output_filename[])
{
size_t length_of_genome;
char * reference_sequence;
int number_of_snps;
int * snp_locations;
int number_of_samples;
int i;
length_of_genome = genome_length(filename);
reference_sequence = (char *) calloc((length_of_genome +1),sizeof(char));
build_reference_sequence(reference_sequence,filename);
number_of_snps = detect_snps(reference_sequence, filename, length_of_genome);
snp_locations = (int *) calloc((number_of_snps+1),sizeof(int));
build_snp_locations(snp_locations, reference_sequence);
free(reference_sequence);
number_of_samples = number_of_sequences_in_file(filename);
// Find out the names of the sequences
char* sequence_names[number_of_samples];
sequence_names[number_of_samples-1] = '\0';
for(i = 0; i < number_of_samples; i++)
{
sequence_names[i] = calloc(MAX_SAMPLE_NAME_SIZE,sizeof(char));
}
get_sample_names_for_header(filename, sequence_names, number_of_samples);
char* bases_for_snps[number_of_snps];
for(i = 0; i < number_of_snps; i++)
{
bases_for_snps[i] = calloc(number_of_samples+1 ,sizeof(char));
}
get_bases_for_each_snp(filename, snp_locations, bases_for_snps, length_of_genome, number_of_snps);
char output_filename_base[FILENAME_MAX];
char filename_without_directory[FILENAME_MAX];
strip_directory_from_filename(filename, filename_without_directory);
strncpy(output_filename_base, filename_without_directory, FILENAME_MAX);
if(output_filename != NULL && *output_filename != '\0')
{
strncpy(output_filename_base, output_filename, FILENAME_MAX);
}
if(output_vcf_file)
{
char vcf_output_filename[FILENAME_MAX];
strncpy(vcf_output_filename, output_filename_base, FILENAME_MAX);
if((output_vcf_file + output_phylip_file + output_multi_fasta_file) > 1 || (output_filename == NULL || *output_filename == '\0') )
{
strcat(vcf_output_filename, ".vcf");
}
create_vcf_file(vcf_output_filename, snp_locations, number_of_snps, bases_for_snps, sequence_names, number_of_samples, length_of_genome);
}
if(output_phylip_file)
{
char phylip_output_filename[FILENAME_MAX];
strncpy(phylip_output_filename, output_filename_base, FILENAME_MAX);
if((output_vcf_file + output_phylip_file + output_multi_fasta_file) > 1 || (output_filename == NULL || *output_filename == '\0') )
{
strcat(phylip_output_filename, ".phylip");
}
create_phylib_of_snp_sites(phylip_output_filename, number_of_snps, bases_for_snps, sequence_names, number_of_samples);
}
if((output_multi_fasta_file) || (output_vcf_file ==0 && output_phylip_file == 0 && output_multi_fasta_file == 0))
{
char multi_fasta_output_filename[FILENAME_MAX];
strncpy(multi_fasta_output_filename, output_filename_base, FILENAME_MAX);
if((output_vcf_file + output_phylip_file + output_multi_fasta_file) > 1 || (output_filename == NULL || *output_filename == '\0') )
{
strcat(multi_fasta_output_filename, ".snp_sites.aln");
}
create_fasta_of_snp_sites(multi_fasta_output_filename, number_of_snps, bases_for_snps, sequence_names, number_of_samples);
}
// free memory
free(snp_locations);
for(i = 0; i < number_of_samples; i++)
{
free(sequence_names[i]);
}
for(i = 0; i < number_of_snps; i++)
{
free(bases_for_snps[i]);
}
return 1;
}