annotate gene_fraction/src/FastaRecord.cpp @ 0:f95150c37d38 draft default tip

planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
author chrisd
date Sun, 21 Feb 2016 23:31:55 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
1 #include "FastaRecord.h"
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
2
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
3 #include <algorithm>
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
4
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
5 FastaRecord::FastaRecord(std::string gene_id, std::string gene) :
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
6 _gene_id(gene_id), _gene(gene), _base_hits(_gene.length(), 0),
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
7 _gene_hits(0) {}
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
8
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
9 std::string FastaRecord::gene_id() const { return _gene_id; }
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
10
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
11 std::string FastaRecord::gene() const { return _gene; }
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
12
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
13 void FastaRecord::update_gene_hits() {
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
14 _gene_hits++;
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
15 }
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
16
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
17 int FastaRecord::gene_hits() const {
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
18 return _gene_hits;
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
19 }
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
20
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
21 int FastaRecord::get_base_hits() const {
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
22 return static_cast<int>(count(_base_hits.begin(), _base_hits.end(), 1));
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
23 }
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
24
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
25 int FastaRecord::find_gene(const std::vector<FastaRecord> &records,
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
26 const std::string &gene_id, std::string seq) {
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
27 int gene_index;
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
28
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
29 std::vector<FastaRecord>::const_iterator low;
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
30 // binary search for fasta record index
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
31 low = std::lower_bound(records.begin(), records.end(), FastaRecord(gene_id, seq),
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
32 [](const FastaRecord &a, const FastaRecord &b)
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
33 { return a.gene_id() < b.gene_id(); });
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
34 gene_index = (low - records.begin());
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
35
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
36 return gene_index;
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
37 }
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
38
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
39 void FastaRecord::sort_by_gene_id(std::vector<FastaRecord> &records) {
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
40 // sort records by gene id
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
41 sort(records.begin(), records.end(), [](const FastaRecord &a, const FastaRecord &b) { return a.gene_id() < b.gene_id(); });
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
42 }
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
43
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
44 void FastaRecord::reset_base_hits(std::vector<FastaRecord> &records) {
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
45 for_each(records.begin(), records.end(), [](FastaRecord &a) { std::fill(a.base_hits().begin(), a.base_hits().end(), 0); });
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
46 }
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
47
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
48 void FastaRecord::reset_gene_hits(std::vector<FastaRecord> &records) {
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
49 for_each(records.begin(), records.end(), [](FastaRecord &a) { a._gene_hits = 0; });
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
50 }
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
51
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
52 std::vector<int> &FastaRecord::base_hits() {
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
53 return _base_hits;
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
54 }
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
55
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
56
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
57
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
58
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
59
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
60
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
61
f95150c37d38 planemo upload for repository https://github.com/ChrisD11/Tools commit ddc95e5d6b5f2c0a5340c0bc384aa822db8856d5
chrisd
parents:
diff changeset
62