void Annotator::annotate_without_map() throw (AnnotatorException) {
Descriptor* descriptor = NULL;
ofstream ofile_stream;
const char* output_prefix = NULL;
const char* file_name = NULL;
char* output_file_name = NULL;
char* line = NULL;
int line_length = 0;
unsigned int line_number = 2u;
char header_separator = '\0';
char data_separator = '\0';
char region_separator = '\0';
bool regions_append = false;
char* token = NULL;
char* end_ptr = NULL;
int column_position = 0;
const char* column_name = NULL;
char* marker_token = NULL;
char* chr_token = NULL;
char* position_token = NULL;
int position = 0;
vector<double>* deviation = NULL;
int deviation_value = 0;
IntervalTree<char*>* genes_index = NULL;
IntervalTree<char*> genes_index_subset;
map<int, vector<char*>*> annotated_genes;
map<int, vector<char*>*>::iterator annotated_genes_it;
if (gwafile == NULL) {
return;
}
try {
descriptor = gwafile->get_descriptor();
output_prefix = descriptor->get_property(Descriptor::PREFIX);
file_name = descriptor->get_name();
header_separator = gwafile->get_header_separator();
data_separator = gwafile->get_data_separator();
deviation = descriptor->get_threshold(Descriptor::REGIONS_DEVIATION);
regions_append = gwafile->is_regions_append_on();
auxiliary::transform_file_name(&output_file_name, output_prefix, file_name, NULL, true);
if (output_file_name == NULL) {
throw AnnotatorException("Annotator", "annotate_without_map()", __LINE__, 16);
}
if (data_separator == ',') {
region_separator = ';';
} else {
region_separator = ',';
}
ofile_stream.exceptions(ios_base::failbit | ios_base::badbit);
try {
ofile_stream.open(output_file_name);
} catch (ofstream::failure &e) {
throw AnnotatorException("Annotator", "annotate_without_map()", __LINE__, 17, output_file_name);
}
try {
if (regions_append) {
ofile_stream << header_backup;
} else {
ofile_stream << ((column_name = descriptor->get_column(Descriptor::MARKER)) != NULL ? column_name : Descriptor::MARKER);
ofile_stream << header_separator;
ofile_stream << ((column_name = descriptor->get_column(Descriptor::CHR)) != NULL ? column_name : Descriptor::CHR);
ofile_stream << header_separator;
ofile_stream << ((column_name = descriptor->get_column(Descriptor::POSITION)) != NULL ? column_name : Descriptor::POSITION);
}
for (unsigned int i = 0u; i < deviation->size(); ++i) {
deviation_value = (int)deviation->at(i);
if (deviation_value != 0) {
ofile_stream << header_separator << "+/-" << deviation_value;
} else {
ofile_stream << header_separator << "IN";
}
}
ofile_stream << endl;
while ((line_length = reader.read_line()) > 0) {
line = *reader.line;
if (regions_append) {
ofile_stream << line;
}
column_position = 0;
marker_token = NULL;
chr_token = NULL;
position_token = NULL;
token = auxiliary::strtok(&line, data_separator);
while (token != NULL) {
if (column_position == marker_column_pos) {
marker_token = token;
} else if (column_position == chr_column_pos) {
chr_token = token;
} else if (column_position == position_column_pos) {
position_token = token;
}
token = auxiliary::strtok(&line, data_separator);
++column_position;
}
if (column_position < total_columns) {
throw AnnotatorException("Annotator", "annotate_without_map()", __LINE__, 8, line_number, gwafile->get_descriptor()->get_name(), column_position, total_columns);
} else if (column_position > total_columns) {
throw AnnotatorException("Annotator", "annotate_without_map()", __LINE__, 9, line_number, gwafile->get_descriptor()->get_name(), column_position, total_columns);
}
position = (int)strtol(position_token, &end_ptr, 10);
if (*end_ptr != '\0') {
throw AnnotatorException("Annotator", "annotate_without_map()", __LINE__, 10, position_token, ((column_name = descriptor->get_column(Descriptor::POSITION)) != NULL) ? column_name : Descriptor::POSITION, line_number);
}
if (position < 0) {
throw AnnotatorException("Annotator", "annotate_without_map()", __LINE__, 11, ((column_name = descriptor->get_column(Descriptor::POSITION)) != NULL) ? column_name : Descriptor::POSITION, position, line_number);
}
if (!regions_append) {
ofile_stream << marker_token << data_separator << chr_token << data_separator << position_token;
}
regions_indices_it = regions_indices.find(chr_token);
if (regions_indices_it != regions_indices.end()) {
genes_index = regions_indices_it->second;
deviation_value = (int)deviation->back();
genes_index->get_intersecting_intervals(position - deviation_value, position + deviation_value, deviation_value, genes_index_subset);
for (int i = deviation->size() - 2; i >= 0; --i) {
deviation_value = (int)deviation->at(i);
genes_index_subset.mark_intersecting_intervals(position - deviation_value, position + deviation_value, deviation_value);
}
genes_index_subset.get_marked_values(annotated_genes);
if (annotated_genes.size() > 0) {
for (unsigned int i = 0u; i < deviation->size(); ++i) {
ofile_stream << data_separator;
annotated_genes_it = annotated_genes.find(((int)deviation->at(i)));
if (annotated_genes_it != annotated_genes.end()) {
write_char_vector(ofile_stream, annotated_genes_it->second, region_separator);
} else {
ofile_stream << "NA";
}
}
ofile_stream << endl;
} else {
for (unsigned int i = 0u; i < deviation->size(); ++i) {
ofile_stream << data_separator << "NA";
}
ofile_stream << endl;
}
annotated_genes_it = annotated_genes.begin();
while(annotated_genes_it != annotated_genes.end()) {
delete annotated_genes_it->second;
annotated_genes_it++;
}
annotated_genes.clear();
genes_index_subset.clear();
} else {
for (unsigned int i = 0u; i < deviation->size(); ++i) {
ofile_stream << data_separator << "NA";
}
ofile_stream << endl;
}
++line_number;
}
} catch (ofstream::failure &e) {
throw AnnotatorException("Annotator", "annotate_without_map()", __LINE__, 19, output_file_name);
}
try {
ofile_stream.close();
} catch (ofstream::failure &e) {
throw AnnotatorException("Annotator", "annotate_without_map()", __LINE__, 18, output_file_name);
}
if (line_length == 0) {
throw AnnotatorException("Annotator", "annotate_without_map()", __LINE__, 13, line_number, gwafile->get_descriptor()->get_name());
}
} catch (DescriptorException &e) {
AnnotatorException new_e(e);
new_e.add_message("Annotator", "annotate_without_map()", __LINE__, 14, gwafile->get_descriptor()->get_name());
throw new_e;
} catch (ReaderException &e) {
AnnotatorException new_e(e);
new_e.add_message("Annotator", "annotate_without_map()", __LINE__, 14, gwafile->get_descriptor()->get_name());
throw new_e;
} catch (AnnotatorException &e) {
e.add_message("Annotator", "annotate_without_map()", __LINE__, 14, gwafile->get_descriptor()->get_name());
throw;
}
}
void Annotator::annotate_with_map() throw (AnnotatorException) {
Descriptor* descriptor = NULL;
ofstream ofile_stream;
const char* output_prefix = NULL;
const char* file_name = NULL;
char* output_file_name = NULL;
char* line = NULL;
int line_length = 0;
unsigned int line_number = 2u;
char* line_backup = NULL;
char header_separator = '\0';
char data_separator = '\0';
char region_separator = '\0';
bool regions_append = false;
char* token = NULL;
int column_position = 0;
const char* column_name = NULL;
marker_index key_marker_index;
marker_index* found_marker_index = NULL;
unsigned int found_marker_index_pos = 0u;
unsigned int index = 0u;
char* chr = NULL;
int position = 0;
vector<double>* deviation = NULL;
int deviation_value = 0;
IntervalTree<char*>* genes_index = NULL;
IntervalTree<char*> genes_index_subset;
map<int, vector<char*>*> annotated_genes;
map<int, vector<char*>*>::iterator annotated_genes_it;
if (gwafile == NULL) {
return;
}
try {
descriptor = gwafile->get_descriptor();
output_prefix = descriptor->get_property(Descriptor::PREFIX);
file_name = descriptor->get_name();
header_separator = gwafile->get_header_separator();
data_separator = gwafile->get_data_separator();
deviation = descriptor->get_threshold(Descriptor::REGIONS_DEVIATION);
regions_append = gwafile->is_regions_append_on();
auxiliary::transform_file_name(&output_file_name, output_prefix, file_name, NULL, true);
if (output_file_name == NULL) {
throw AnnotatorException("Annotator", "annotate_with_map()", __LINE__, 16);
}
if (regions_append) {
line_backup = (char*)malloc(reader.get_buffer_size() * sizeof(char));
if (line_backup == NULL) {
throw AnnotatorException("Annotator", "annotate_with_map()", __LINE__, 2, (reader.get_buffer_size() * sizeof(char)));
}
}
if (data_separator == ',') {
region_separator = ';';
} else {
region_separator = ',';
}
ofile_stream.exceptions(ios_base::failbit | ios_base::badbit);
try {
ofile_stream.open(output_file_name);
} catch (ofstream::failure &e) {
throw AnnotatorException("Annotator", "annotate_with_map()", __LINE__, 17, output_file_name);
}
try {
if (regions_append) {
ofile_stream << header_backup;
ofile_stream << header_separator;
ofile_stream << ((column_name = descriptor->get_property(Descriptor::MAP_CHR)) != NULL ? column_name : Descriptor::MAP_CHR);
ofile_stream << header_separator;
ofile_stream << ((column_name = descriptor->get_property(Descriptor::MAP_POSITION)) != NULL ? column_name : Descriptor::MAP_POSITION);
} else {
ofile_stream << ((column_name = descriptor->get_column(Descriptor::MARKER)) != NULL ? column_name : Descriptor::MARKER);
ofile_stream << header_separator;
ofile_stream << ((column_name = descriptor->get_property(Descriptor::MAP_CHR)) != NULL ? column_name : Descriptor::MAP_CHR);
ofile_stream << header_separator;
ofile_stream << ((column_name = descriptor->get_property(Descriptor::MAP_POSITION)) != NULL ? column_name : Descriptor::MAP_POSITION);
}
for (unsigned int i = 0u; i < deviation->size(); ++i) {
deviation_value = (int)deviation->at(i);
if (deviation_value != 0) {
ofile_stream << header_separator << "+/-" << deviation_value;
} else {
ofile_stream << header_separator << "IN";
}
}
ofile_stream << endl;
while ((line_length = reader.read_line()) > 0) {
line = *reader.line;
if (regions_append) {
strcpy(line_backup, line);
}
column_position = 0;
key_marker_index.name = NULL;
token = auxiliary::strtok(&line, data_separator);
while (token != NULL) {
if (column_position == marker_column_pos) {
key_marker_index.name = token;
}
token = auxiliary::strtok(&line, data_separator);
++column_position;
}
if (column_position < total_columns) {
throw AnnotatorException("Annotator", "annotate_with_map()", __LINE__, 8, line_number, gwafile->get_descriptor()->get_name(), column_position, total_columns);
} else if (column_position > total_columns) {
throw AnnotatorException("Annotator", "annotate_with_map()", __LINE__, 9, line_number, gwafile->get_descriptor()->get_name(), column_position, total_columns);
}
found_marker_index = (marker_index*)bsearch(&key_marker_index, map_index, map_index_size, sizeof(marker_index), qsort_marker_index_cmp);
if (found_marker_index == NULL) {
if (regions_append) {
ofile_stream << line_backup << data_separator << "NA" << data_separator << "NA";
} else {
ofile_stream << key_marker_index.name << data_separator << "NA" << data_separator << "NA";
}
for (unsigned int i = 0u; i < deviation->size(); ++i) {
ofile_stream << data_separator << "NA";
}
ofile_stream << endl;
} else {
found_marker_index_pos = found_marker_index - map_index;
while ((found_marker_index_pos < map_index_size) && (auxiliary::strcmp_ignore_case(key_marker_index.name, map_index[found_marker_index_pos].name) == 0)) {
index = map_index[found_marker_index_pos].index;
chr = map_chromosomes[index];
position = map_positions[index];
if (regions_append) {
ofile_stream << line_backup << data_separator << chr << data_separator << position;
} else {
ofile_stream << key_marker_index.name << data_separator << chr << data_separator << position;
}
regions_indices_it = regions_indices.find(chr);
if (regions_indices_it != regions_indices.end()) {
genes_index = regions_indices_it->second;
deviation_value = (int)deviation->back();
genes_index->get_intersecting_intervals(position - deviation_value, position + deviation_value, deviation_value, genes_index_subset);
for (int i = deviation->size() - 2; i >= 0; --i) {
deviation_value = (int)deviation->at(i);
genes_index_subset.mark_intersecting_intervals(position - deviation_value, position + deviation_value, deviation_value);
}
genes_index_subset.get_marked_values(annotated_genes);
if (annotated_genes.size() > 0) {
for (unsigned int i = 0u; i < deviation->size(); ++i) {
ofile_stream << data_separator;
annotated_genes_it = annotated_genes.find(((int)deviation->at(i)));
if (annotated_genes_it != annotated_genes.end()) {
write_char_vector(ofile_stream, annotated_genes_it->second, region_separator);
} else {
ofile_stream << "NA";
}
}
ofile_stream << endl;
} else {
for (unsigned int i = 0u; i < deviation->size(); ++i) {
ofile_stream << data_separator << "NA";
}
ofile_stream << endl;
}
annotated_genes_it = annotated_genes.begin();
while(annotated_genes_it != annotated_genes.end()) {
delete annotated_genes_it->second;
annotated_genes_it++;
}
annotated_genes.clear();
genes_index_subset.clear();
} else {
for (unsigned int i = 0u; i < deviation->size(); ++i) {
ofile_stream << data_separator << "NA";
}
ofile_stream << endl;
}
++found_marker_index_pos;
}
}
++line_number;
}
} catch (ofstream::failure &e) {
throw AnnotatorException("Annotator", "annotate_with_map()", __LINE__, 19, output_file_name);
}
try {
ofile_stream.close();
} catch (ofstream::failure &e) {
throw AnnotatorException("Annotator", "annotate_with_map()", __LINE__, 18, output_file_name);
}
if (line_backup != NULL) {
free(line_backup);
line_backup = NULL;
}
if (line_length == 0) {
throw AnnotatorException("Annotator", "annotate_with_map()", __LINE__, 13, line_number, gwafile->get_descriptor()->get_name());
}
} catch (DescriptorException &e) {
AnnotatorException new_e(e);
new_e.add_message("Annotator", "annotate_with_map()", __LINE__, 14, gwafile->get_descriptor()->get_name());
throw new_e;
} catch (ReaderException &e) {
AnnotatorException new_e(e);
new_e.add_message("Annotator", "annotate_with_map()", __LINE__, 14, gwafile->get_descriptor()->get_name());
throw new_e;
} catch (AnnotatorException &e) {
e.add_message("Annotator", "annotate_with_map()", __LINE__, 14, gwafile->get_descriptor()->get_name());
throw;
}
}
void Formatter::format(double lambda, char new_separator, int& n_total, int& n_filtered) throw (FormatterException) {
Descriptor* descriptor = NULL;
char* line = NULL;
int line_length = 0;
unsigned int line_number = 2;
char* token = NULL;
char data_separator = '\0';
const char* output_prefix = NULL;
const char* file_name = NULL;
char* o_gwafile_name = NULL;
vector<double>* snp_hq = NULL;
double maf_filter_value = -numeric_limits<double>::infinity();
double oevar_imp_filter_value = -numeric_limits<double>::infinity();
bool maf_filter = false;
bool oevar_imp_filter = false;
Column* column = NULL;
ofstream ofile_stream;
double d_value = 0.0;
n_total = 0;
n_filtered = 0;
if (gwafile == NULL) {
return;
}
try {
descriptor = gwafile->get_descriptor();
output_prefix = descriptor->get_property(Descriptor::PREFIX);
file_name = descriptor->get_name();
snp_hq = descriptor->get_threshold(Descriptor::SNP_HQ);
data_separator = gwafile->get_data_separator();
auxiliary::transform_file_name(&o_gwafile_name, output_prefix, file_name, NULL, true);
if (o_gwafile_name == NULL) {
throw FormatterException("Formatter", "format( double , char , int& , int& )", __LINE__, 10);
}
if (((maf_filter_value = snp_hq->at(0)) > 0) && (maf_column_pos >= 0)){
maf_filter = true;
}
if (((oevar_imp_filter_value = snp_hq->at(1)) > 0) && (oevar_imp_column_pos >= 0)) {
oevar_imp_filter = true;
}
for (columns_it = input_columns.begin(); columns_it != input_columns.end(); columns_it++) {
output_columns.push_back(*columns_it);
}
if (!isnan(lambda)) {
if (stderr_column != NULL) {
column = new CorrectedStandardErrorColumn(stderr_column, lambda);
column->set_header("%s_gc", stderr_column->get_header());
column->set_order(descriptor->get_column_order(column->get_header(), gwafile->is_case_sensitive()));
output_columns.push_back(column);
}
if (pvalue_column != NULL) {
column = new CorrectedPvalueColumn(pvalue_column, lambda);
column->set_header("%s_gc", pvalue_column->get_header());
column->set_order(descriptor->get_column_order(column->get_header(), gwafile->is_case_sensitive()));
output_columns.push_back(column);
}
}
if ((oevar_imp_column != NULL) && (n_total_column != NULL)) {
column = new EffectiveSampleSizeColumn(n_total_column, oevar_imp_column);
column->set_header("%s_effective", n_total_column->get_header());
column->set_order(descriptor->get_column_order(column->get_header(), gwafile->is_case_sensitive()));
output_columns.push_back(column);
}
if (gwafile->is_order_on()) {
if (descriptor->get_reordered_columns_number() > 0) {
stable_sort(output_columns.begin(), output_columns.end(), compare_columns);
} else {
stable_sort(output_columns.begin(), output_columns.end(), compare_columns_by_name);
}
}
ofile_stream.exceptions(ios_base::failbit | ios_base::badbit);
ofile_stream.precision(numeric_limits<double>::digits10);
try {
ofile_stream.open(o_gwafile_name);
} catch (ofstream::failure &e) {
throw FormatterException("Formatter", "format( double , char , int& , int& )", __LINE__, 11, o_gwafile_name);
}
try {
columns_it = output_columns.begin();
if (columns_it != output_columns.end()) {
ofile_stream << (*columns_it)->get_header();
while (++columns_it != output_columns.end()) {
ofile_stream << new_separator << (*columns_it)->get_header();
}
ofile_stream << endl;
}
if (maf_filter) {
if (oevar_imp_filter) {
// all filters
while ((line_length = reader.read_line()) > 0) {
line = *reader.line;
// for (columns_it = input_columns.begin(); columns_it != input_columns.end(); (*columns_it)->char_value = auxiliary::strtok(&line, data_separator), columns_it++);
for (columns_it = input_columns.begin(); columns_it != input_columns.end(); ++columns_it) {
token = auxiliary::strtok(&line, data_separator);
if (token == NULL) {
(*columns_it)->char_value = "";
} else {
auxiliary::trim(&token);
(*columns_it)->char_value = token;
}
// auxiliary::trim(&token);
// (*columns_it)->char_value = token;
}
d_value = maf_column->get_numeric_value();
d_value = d_value > 0.5 ? 1.0 - d_value : d_value;
if (auxiliary::fcmp(d_value, maf_filter_value, EPSILON) == 1) {
d_value = oevar_imp_column->get_numeric_value();
if (auxiliary::fcmp(d_value, oevar_imp_filter_value, EPSILON) == 1) {
columns_it = output_columns.begin();
if (columns_it != output_columns.end()) {
(*columns_it)->out(ofile_stream);
while (++columns_it != output_columns.end()) {
ofile_stream << new_separator;
(*columns_it)->out(ofile_stream);
}
ofile_stream << endl;
}
n_filtered += 1;
}
}
line_number += 1;
}
} else {
// only maf filter
while ((line_length = reader.read_line()) > 0) {
line = *reader.line;
// for (columns_it = input_columns.begin(); columns_it != input_columns.end(); (*columns_it)->char_value = auxiliary::strtok(&line, data_separator), columns_it++);
for (columns_it = input_columns.begin(); columns_it != input_columns.end(); ++columns_it) {
token = auxiliary::strtok(&line, data_separator);
if (token == NULL) {
(*columns_it)->char_value = "";
} else {
auxiliary::trim(&token);
(*columns_it)->char_value = token;
}
// auxiliary::trim(&token);
// (*columns_it)->char_value = token;
}
d_value = maf_column->get_numeric_value();
d_value = d_value > 0.5 ? 1.0 - d_value : d_value;
if (auxiliary::fcmp(d_value, maf_filter_value, EPSILON) == 1) {
columns_it = output_columns.begin();
if (columns_it != output_columns.end()) {
(*columns_it)->out(ofile_stream);
while (++columns_it != output_columns.end()) {
ofile_stream << new_separator;
(*columns_it)->out(ofile_stream);
}
ofile_stream << endl;
}
n_filtered += 1;
}
line_number += 1;
}
}
} else if (oevar_imp_filter) {
// only oevar_imp filter
while ((line_length = reader.read_line()) > 0) {
line = *reader.line;
// for (columns_it = input_columns.begin(); columns_it != input_columns.end(); (*columns_it)->char_value = auxiliary::strtok(&line, data_separator), columns_it++);
for (columns_it = input_columns.begin(); columns_it != input_columns.end(); ++columns_it) {
token = auxiliary::strtok(&line, data_separator);
if (token == NULL) {
(*columns_it)->char_value = "";
} else {
auxiliary::trim(&token);
(*columns_it)->char_value = token;
}
// auxiliary::trim(&token);
// (*columns_it)->char_value = token;
}
d_value = oevar_imp_column->get_numeric_value();
if (auxiliary::fcmp(d_value, oevar_imp_filter_value, EPSILON) == 1) {
columns_it = output_columns.begin();
if (columns_it != output_columns.end()) {
(*columns_it)->out(ofile_stream);
while (++columns_it != output_columns.end()) {
ofile_stream << new_separator;
(*columns_it)->out(ofile_stream);
}
ofile_stream << endl;
}
n_filtered += 1;
}
line_number += 1;
}
} else {
// no filters
while ((line_length = reader.read_line()) > 0) {
line = *reader.line;
// for (columns_it = input_columns.begin(); columns_it != input_columns.end(); (*columns_it)->char_value = auxiliary::strtok(&line, data_separator), columns_it++);
for (columns_it = input_columns.begin(); columns_it != input_columns.end(); ++columns_it) {
token = auxiliary::strtok(&line, data_separator);
if (token == NULL) {
(*columns_it)->char_value = "";
} else {
auxiliary::trim(&token);
(*columns_it)->char_value = token;
}
// auxiliary::trim(&token);
// (*columns_it)->char_value = token;
}
columns_it = output_columns.begin();
if (columns_it != output_columns.end()) {
(*columns_it)->out(ofile_stream);
while (++columns_it != output_columns.end()) {
ofile_stream << new_separator;
(*columns_it)->out(ofile_stream);
}
ofile_stream << endl;
}
n_filtered += 1;
line_number += 1;
}
}
} catch (ofstream::failure &e) {
throw FormatterException("Formatter", "format( double , char , int& , int& )", __LINE__, 13, o_gwafile_name);
}
try {
ofile_stream.close();
} catch (ofstream::failure &e) {
throw FormatterException("Formatter", "format( double , char , int& , int& )", __LINE__, 12, o_gwafile_name);
}
n_total = line_number - 2;
if (line_length == 0) {
throw FormatterException("Formatter", "format( double , char , int& , int& )", __LINE__, 7, line_number, gwafile->get_descriptor()->get_name());
}
} catch (DescriptorException &e) {
FormatterException new_e(e);
e.add_message("Formatter", "format( double , char , int& , int& )", __LINE__, 14, gwafile->get_descriptor()->get_name());
throw new_e;
} catch (ColumnException &e) {
FormatterException new_e(e);
e.add_message("Formatter", "format( double , char , int& , int& )", __LINE__, 14, gwafile->get_descriptor()->get_name());
throw new_e;
} catch (FormatterException &e) {
e.add_message("Formatter", "format( double , char , int& , int& )", __LINE__, 14, gwafile->get_descriptor()->get_name());
throw;
}
}
