Mercurial > repos > chrisd > testshed

annotate gene_fraction/src/SamRatio.h @ 0:0fd352f62446 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
author chrisd
date Sun, 21 Feb 2016 06:05:24 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
1 #ifndef SAM_RATIO_H
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
2 #define SAM_RATIO_H
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
3
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
4 #include <iostream>
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
5 #include <fstream>
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
6 #include <algorithm>
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
7
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
8 #include "FastaRecord.h"
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
9 #include "Alignments.h"
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
10 #include "args.h"
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
11
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
12 typedef std::vector<std::pair<int,char>> cigar_str;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
13
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
14 /**
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
15 *
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
16 */
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
17 struct header {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
18 std::string level = "Level";
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
19 std::string iteration = "Iteration";
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
20 std::string gene_id = "Gene id";
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
21 std::string gene_fraction = "Gene Fraction";
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
22 std::string hits = "Hits";
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
23 };
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
24
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
25 /**
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
26 * Reports the total number of bases that were touched for each
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
27 * gene. This is largely calculated using the positional and seq
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
28 * information found in fields four and ten of each alignment
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
29 */
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
30 void analyze_coverage(FastaRecord &record, Alignments &alignment) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
31 record.update_gene_hits();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
32
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
33 cigar_str cigar = alignment.cigar();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
34
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
35 int len;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
36 char op;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
37
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
38 int occurrence;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
39 int pos_in_gene = alignment.pos();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
40
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
41 int start, stop;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
42 int base_hits = record._base_hits.size(); // func this
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
43 int read_length = alignment.seq().length(); //func this
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
44
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
45 if(pos_in_gene == 1) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
46 occurrence = 0;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
47 for(int i = 0; i < cigar.size(); i++) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
48 len = cigar[i].first;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
49 op = cigar[i].second;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
50
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
51 switch(op) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
52 case 'M':
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
53 occurrence += len;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
54 break;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
55 case 'I':
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
56 occurrence += len;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
57 break;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
58 default:
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
59 break;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
60 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
61 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
62
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
63 start = read_length - occurrence;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
64 stop = start + read_length;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
65
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
66 for(int i = start; i < base_hits; i++) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
67 if(i == stop) break;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
68 record._base_hits[i] = 1;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
69 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
70 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
71 else {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
72 start = pos_in_gene - 1;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
73 stop = start + read_length;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
74
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
75 for(int i = start; i < base_hits; i++) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
76 if(i == stop) break;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
77 record._base_hits[i] = 1;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
78 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
79 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
80 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
81
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
82 /**
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
83 * Returns gene fraction of fasta record
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
84 * Returns -1 if gene fraction is not greater than threshold
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
85 */
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
86 double coverage(const FastaRecord &record, const int &threshold) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
87 double gene_coverage;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
88
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
89 int base_hits, gene_size;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
90
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
91 base_hits = record.get_base_hits();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
92 gene_size = record.gene().length();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
93
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
94 gene_coverage = (static_cast<double>(base_hits)/static_cast<double>(gene_size))*100;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
95
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
96 if(gene_coverage > threshold)
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
97 return gene_coverage;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
98 return -1;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
99 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
100
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
101 /**
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
102 * Writes header to output file when
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
103 * reading from stdin
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
104 */
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
105 void bam_stream_header() {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
106 header h;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
107 char sep = ',';
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
108
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
109 std::cout << h.level << sep << h.iteration << sep
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
110 << h.gene_id << sep << h.gene_fraction << sep
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
111 << h.hits << '\n';
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
112 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
113
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
114 /**
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
115 * Writes header to output file when
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
116 * reading from sam file
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
117 */
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
118 void file_header(const std::string &out_fp, const std::string &sam_fp) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
119 header h;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
120 std::ofstream out(out_fp.c_str(), std::ofstream::app );
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
121 char sep = ',';
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
122
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
123 out << "@" << sam_fp << '\n';
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
124 out << h.level << sep << h.iteration << sep
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
125 << h.gene_id << sep << h.gene_fraction << sep
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
126 << h.hits << '\n';
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
127 out.close();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
128 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
129
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
130 /**
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
131 *
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
132 */
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
133 void create_header(cmd_args &args) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
134 if(args.bam_stream) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
135 bam_stream_header();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
136 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
137 else {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
138 file_header(args.out_fp, args.sam_fp);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
139 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
140 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
141
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
142 /**
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
143 * Writes results to output file when reading from
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
144 * stdin
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
145 */
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
146 void bam_stream_results(std::vector<FastaRecord> &records,
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
147 const int &level, const int &iteration,
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
148 cmd_args &args) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
149
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
150 double gene_fraction;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
151 int hits_seen;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
152 std::string id;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
153 char sep = ',';
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
154
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
155 for(auto &rec : records) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
156 gene_fraction = coverage(rec, args.threshold);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
157 hits_seen = rec.gene_hits();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
158 id = rec.gene_id();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
159
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
160 if(gene_fraction > 0) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
161 std::cout << level << sep << iteration << sep
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
162 << id << sep << gene_fraction << sep
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
163 << hits_seen << '\n';
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
164 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
165 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
166 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
167
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
168 /**
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
169 * Write results when reading sam file from
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
170 * path
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
171 */
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
172 void file_results(std::vector<FastaRecord> &records,
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
173 const int level, const int &iteration,
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
174 cmd_args &args) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
175
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
176 std::ofstream out(args.out_fp.c_str(), std::ofstream::app);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
177
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
178 double gene_fraction;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
179 int hits_seen;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
180 std::string id;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
181 char sep = ',';
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
182
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
183 for(auto &rec : records) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
184 gene_fraction = coverage(rec, args.threshold);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
185 hits_seen = rec.gene_hits();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
186 id = rec.gene_id();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
187
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
188 if(gene_fraction > 0) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
189 out << level << sep << iteration << sep
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
190 << id << sep << gene_fraction << sep
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
191 << hits_seen << '\n';
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
192 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
193 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
194 out.close();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
195 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
196
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
197 /**
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
198 *
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
199 */
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
200 void report_results(std::vector<FastaRecord> &records,
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
201 const int level, const int &iteration,
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
202 cmd_args &args) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
203
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
204 if(args.bam_stream) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
205 bam_stream_results(records,level,iteration,args);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
206 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
207 else {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
208 file_results(records,level,iteration,args);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
209 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
210 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
211
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
212 /**
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
213 * Generates a sequence of samples from sam file specified
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
214 * by the starting level, max level, and skip pattern
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
215 */
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
216 void generate_samples(std::vector<FastaRecord> &records,
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
217 std::vector<Alignments> &alignments,
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
218 cmd_args &args) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
219
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
220 int read_count = alignments.size();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
221 int sample_size;
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
222
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
223 srand(unsigned(time(0)));
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
224
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
225 std::vector<int> sequence(read_count);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
226 iota(sequence.begin(), sequence.end(), 0);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
227
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
228 create_header(args);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
229
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
230 for(int level = args.min; level <= args.max; level += args.skip) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
231 for(int sample = 0; sample < args.s_per_lev; sample++) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
232 random_shuffle(sequence.begin(), sequence.end(), randomize);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
233 sample_size = round(((static_cast<double>(level)/100)*read_count));
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
234 std::vector<int> chosen(sequence.begin(), sequence.begin()+sample_size);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
235
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
236 for(int a_idx = 0; a_idx < chosen.size(); a_idx++) {
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
237 std::string rname = alignments[chosen[a_idx]].rname();
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
238 int gene_idx = FastaRecord::find_gene(records, rname);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
239 analyze_coverage(records[gene_idx], alignments[chosen[a_idx]]);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
240 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
241 report_results(records,level,sample+1,args);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
242 FastaRecord::reset_base_hits(records);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
243 FastaRecord::reset_gene_hits(records);
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
244 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
245 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
246 }
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
247
0fd352f62446 planemo upload for repository https://github.com/ChrisD11/Duplicon commit 3ee0594c692faac542ffa58f4339d79b9b8aefbd-dirty
chrisd
parents:
diff changeset
248 #endif /* SAM_RATIO_H */