Skip to content

Commit 2bb6698

Browse files
Fixing indentation
1 parent f2adba7 commit 2bb6698

File tree

2 files changed

+71
-84
lines changed

2 files changed

+71
-84
lines changed

chromosome_utils.cpp

Lines changed: 50 additions & 49 deletions
Original file line numberDiff line numberDiff line change
@@ -25,6 +25,7 @@
2525
using std::vector;
2626
using std::string;
2727
using std::unordered_map;
28+
using std::runtime_error;
2829

2930
static const char *digits = "0987654321";
3031
static const char *whitespace = " \t";
@@ -33,74 +34,74 @@ static const char *whitespace = " \t";
3334
// thing chrom?
3435

3536
void
36-
parse_region_name(string region_name,
37-
string& chrom, size_t &start, size_t &end) {
38-
37+
parse_region_name(string region_name,
38+
string& chrom, size_t &start, size_t &end) {
39+
3940
const size_t colon_offset = region_name.find(":");
40-
41+
4142
// get the chromosome
4243
size_t chr_offset = region_name.find_last_of(whitespace, colon_offset);
4344
if (chr_offset == string::npos)
4445
chr_offset = 0;
4546
else
4647
chr_offset += 1;
4748
chrom = region_name.substr(chr_offset, colon_offset - chr_offset);
48-
49+
4950
// get the start
5051
const size_t start_end = region_name.find("-", colon_offset + 1);
5152
const string start_string = region_name.substr(colon_offset + 1,
52-
start_end - colon_offset + 1);
53+
start_end - colon_offset + 1);
5354
start = static_cast<size_t>(atoi(start_string.c_str()));
54-
55+
5556
// get the end
5657
const size_t end_end =
5758
region_name.find_first_not_of(digits, start_end + 1);
5859
const string end_string = region_name.substr(start_end + 1,
59-
end_end - start_end - 1);
60+
end_end - start_end - 1);
6061
end = static_cast<size_t>(atoi(end_string.c_str()));
6162
}
6263

6364

64-
static size_t
65+
static size_t
6566
adjust_start_pos(const size_t orig_start, const string &chrom_name) {
6667
static const double LINE_WIDTH = 50.0;
6768
const size_t name_offset = chrom_name.length() + 2; // For the '>' and '\n';
68-
const size_t preceding_newlines =
69+
const size_t preceding_newlines =
6970
static_cast<size_t>(std::floor(orig_start / LINE_WIDTH));
7071
return orig_start + preceding_newlines + name_offset;
7172
}
7273

7374

74-
static size_t
75+
static size_t
7576
adjust_region_size(const size_t orig_start,
76-
const string &chrom_name,
77-
const size_t orig_size) {
77+
const string &chrom_name,
78+
const size_t orig_size) {
7879
static const double LINE_WIDTH = 50.0;
79-
const size_t preceding_newlines_start =
80+
const size_t preceding_newlines_start =
8081
static_cast<size_t>(std::floor(orig_start / LINE_WIDTH));
81-
const size_t preceding_newlines_end =
82+
const size_t preceding_newlines_end =
8283
static_cast<size_t>(std::floor((orig_start + orig_size) / LINE_WIDTH));
8384
return (orig_size + (preceding_newlines_end - preceding_newlines_start));
8485
}
8586

8687

87-
void
88+
void
8889
extract_regions_chrom_fasta(const string &chrom_name,
89-
const string &filename,
90-
const vector<SimpleGenomicRegion> &regions,
91-
vector<string> &sequences) {
92-
90+
const string &filename,
91+
const vector<SimpleGenomicRegion> &regions,
92+
vector<string> &sequences) {
93+
9394
std::ifstream in(filename.c_str());
9495
for (vector<SimpleGenomicRegion>::const_iterator i(regions.begin());
9596
i != regions.end(); ++i) {
96-
97+
9798
const size_t orig_start_pos = i->get_start();
9899
const size_t orig_end_pos = i->get_end();
99100
const size_t orig_region_size = orig_end_pos - orig_start_pos;
100101

101102
const size_t start_pos = adjust_start_pos(orig_start_pos, chrom_name);
102103
const size_t region_size = adjust_region_size(
103-
orig_start_pos, chrom_name, orig_region_size);
104+
orig_start_pos, chrom_name, orig_region_size);
104105
assert(start_pos >= 0);
105106

106107
in.seekg(start_pos);
@@ -109,12 +110,12 @@ extract_regions_chrom_fasta(const string &chrom_name,
109110
in.read(buffer, region_size);
110111

111112
std::remove_if(buffer, buffer + region_size,
112-
std::bind2nd(std::equal_to<char>(), '\n'));
113+
std::bind2nd(std::equal_to<char>(), '\n'));
113114
buffer[orig_region_size] = '\0';
114-
115+
115116
sequences.push_back(buffer);
116117
std::transform(sequences.back().begin(), sequences.back().end(),
117-
sequences.back().begin(), std::ptr_fun(&toupper));
118+
sequences.back().begin(), std::ptr_fun(&toupper));
118119
assert(i->get_width() == sequences.back().length());
119120
}
120121
in.close();
@@ -123,21 +124,21 @@ extract_regions_chrom_fasta(const string &chrom_name,
123124

124125
void
125126
extract_regions_chrom_fasta(const string &chrom_name,
126-
const string &filename,
127-
const vector<GenomicRegion> &regions,
128-
vector<string> &sequences) {
129-
127+
const string &filename,
128+
const vector<GenomicRegion> &regions,
129+
vector<string> &sequences) {
130+
130131
std::ifstream in(filename.c_str());
131132
for (vector<GenomicRegion>::const_iterator i(regions.begin());
132133
i != regions.end(); ++i) {
133-
134+
134135
const size_t orig_start_pos = i->get_start();
135136
const size_t orig_end_pos = i->get_end();
136137
const size_t orig_region_size = orig_end_pos - orig_start_pos;
137138

138139
const size_t start_pos = adjust_start_pos(orig_start_pos, chrom_name);
139140
const size_t region_size = adjust_region_size(
140-
orig_start_pos, chrom_name, orig_region_size);
141+
orig_start_pos, chrom_name, orig_region_size);
141142
assert(start_pos >= 0);
142143

143144
in.seekg(start_pos);
@@ -146,14 +147,14 @@ extract_regions_chrom_fasta(const string &chrom_name,
146147
in.read(buffer, region_size);
147148

148149
std::remove_if(
149-
buffer, buffer + region_size,
150-
std::bind2nd(std::equal_to<char>(), '\n'));
150+
buffer, buffer + region_size,
151+
std::bind2nd(std::equal_to<char>(), '\n'));
151152
buffer[orig_region_size] = '\0';
152153

153154
sequences.push_back(buffer);
154155
std::transform(
155-
sequences.back().begin(), sequences.back().end(),
156-
sequences.back().begin(), std::ptr_fun(&toupper));
156+
sequences.back().begin(), sequences.back().end(),
157+
sequences.back().begin(), std::ptr_fun(&toupper));
157158
if (i->neg_strand())
158159
revcomp_inplace(sequences.back());
159160
assert(i->get_width() == sequences.back().length());
@@ -164,12 +165,12 @@ extract_regions_chrom_fasta(const string &chrom_name,
164165

165166
void
166167
extract_regions_fasta(const string &dirname,
167-
const vector<GenomicRegion> &regions_in,
168-
vector<string> &sequences) {
169-
168+
const vector<GenomicRegion> &regions_in,
169+
vector<string> &sequences) {
170+
170171
static const string FASTA_SUFFIX(".fa");
171172
assert(check_sorted(regions_in));
172-
173+
173174
vector<string> filenames;
174175
read_dir(dirname, filenames);
175176

@@ -187,15 +188,15 @@ extract_regions_fasta(const string &dirname,
187188
std::unordered_map<string, size_t>::const_iterator f_idx =
188189
chrom_regions_map.find(chrom_file);
189190
if (f_idx == chrom_regions_map.end())
190-
throw SMITHLABException("chrom not found:\t" + chrom_file);
191+
throw runtime_error("chrom not found:\t" + chrom_file);
191192
extract_regions_chrom_fasta(
192-
chrom_name, filenames[f_idx->second], regions[i], sequences);
193+
chrom_name, filenames[f_idx->second], regions[i], sequences);
193194
}
194195
}
195196

196197

197198
void extract_regions_fasta(const string &dirname,
198-
const vector<SimpleGenomicRegion> &regions_in, vector<string> &sequences) {
199+
const vector<SimpleGenomicRegion> &regions_in, vector<string> &sequences) {
199200

200201
static const string FASTA_SUFFIX(".fa");
201202
assert(check_sorted(regions_in));
@@ -217,16 +218,16 @@ void extract_regions_fasta(const string &dirname,
217218
std::unordered_map<string, size_t>::const_iterator f_idx =
218219
chrom_regions_map.find(chrom_file);
219220
if (f_idx == chrom_regions_map.end())
220-
throw SMITHLABException("chrom not found:\t" + chrom_file);
221+
throw runtime_error("chrom not found:\t" + chrom_file);
221222
extract_regions_chrom_fasta(
222-
chrom_name, filenames[f_idx->second], regions[i], sequences);
223+
chrom_name, filenames[f_idx->second], regions[i], sequences);
223224
}
224225
}
225226

226227

227228
void
228-
identify_chromosomes(const string chrom_file, const string fasta_suffix,
229-
unordered_map<string, string> &chrom_files) {
229+
identify_chromosomes(const string chrom_file, const string fasta_suffix,
230+
unordered_map<string, string> &chrom_files) {
230231
vector<string> the_files;
231232
if (isdir(chrom_file.c_str())) {
232233
read_dir(chrom_file, fasta_suffix, the_files);
@@ -238,9 +239,9 @@ identify_chromosomes(const string chrom_file, const string fasta_suffix,
238239

239240

240241
void
241-
identify_and_read_chromosomes(const string chrom_file,
242-
const string fasta_suffix,
243-
unordered_map<string, string> &chrom_files) {
242+
identify_and_read_chromosomes(const string chrom_file,
243+
const string fasta_suffix,
244+
unordered_map<string, string> &chrom_files) {
244245
vector<string> the_files;
245246
if (isdir(chrom_file.c_str())) {
246247
read_dir(chrom_file, fasta_suffix, the_files);

chromosome_utils.hpp

Lines changed: 21 additions & 35 deletions
Original file line numberDiff line numberDiff line change
@@ -38,60 +38,46 @@
3838

3939
void
4040
parse_region_name(std::string region_name,
41-
std::string &chrom, size_t &start, size_t &end);
42-
43-
44-
// template <class T> std::string
45-
// assemble_region_name(const T &region) {
46-
// return (region.get_chrom() + ":" + smithlab::toa(region.get_start()) + "-" +
47-
// smithlab::toa(region.get_end()));
48-
// }
49-
50-
51-
// template <class T> std::string
52-
// assemble_region_name(const T &region, const std::string sep) {
53-
// return (region.get_chrom() + sep + smithlab::toa(region.get_start()) + sep +
54-
// smithlab::toa(region.get_end()));
55-
// }
41+
std::string &chrom, size_t &start, size_t &end);
5642

5743

5844
void
59-
extract_regions_chrom_fasta(const std::string &chrom_name,
60-
const std::string &filename,
61-
const std::vector<GenomicRegion> &regions,
62-
std::vector<std::string> &sequences);
45+
extract_regions_chrom_fasta(const std::string &chrom_name,
46+
const std::string &filename,
47+
const std::vector<GenomicRegion> &regions,
48+
std::vector<std::string> &sequences);
6349

6450

6551
void
66-
extract_regions_chrom_fasta(const std::string &chrom_name,
67-
const std::string &filename,
68-
const std::vector<SimpleGenomicRegion> &regions,
69-
std::vector<std::string> &sequences);
52+
extract_regions_chrom_fasta(const std::string &chrom_name,
53+
const std::string &filename,
54+
const std::vector<SimpleGenomicRegion> &regions,
55+
std::vector<std::string> &sequences);
7056

7157

7258
void
73-
extract_regions_fasta(const std::string &dirname,
74-
const std::vector<SimpleGenomicRegion> &regions_in,
75-
std::vector<std::string> &sequences);
59+
extract_regions_fasta(const std::string &dirname,
60+
const std::vector<SimpleGenomicRegion> &regions_in,
61+
std::vector<std::string> &sequences);
7662

7763

7864
void
79-
extract_regions_fasta(const std::string &dirname,
80-
const std::vector<GenomicRegion> &regions_in,
81-
std::vector<std::string> &sequences);
65+
extract_regions_fasta(const std::string &dirname,
66+
const std::vector<GenomicRegion> &regions_in,
67+
std::vector<std::string> &sequences);
8268

8369

8470
void
8571
identify_chromosomes(const std::string chrom_file,
86-
const std::string fasta_suffix,
87-
std::unordered_map<std::string,
88-
std::string> &chrom_files);
72+
const std::string fasta_suffix,
73+
std::unordered_map<std::string,
74+
std::string> &chrom_files);
8975

9076

9177
void
9278
identify_and_read_chromosomes(const std::string chrom_file,
93-
const std::string fasta_suffix,
94-
std::unordered_map<std::string,
95-
std::string> &chrom_files);
79+
const std::string fasta_suffix,
80+
std::unordered_map<std::string,
81+
std::string> &chrom_files);
9682

9783
#endif

0 commit comments

Comments
 (0)