|
| 1 | +#include<iostream> |
| 2 | +#include "../src/trim_primer_quality.h" |
| 3 | +#include "../src/primer_bed.h" |
| 4 | +#include "htslib/sam.h" |
| 5 | + |
| 6 | +int main(){ |
| 7 | + int success = 0; |
| 8 | + std::string bam = "../data/test.sim.merged.sorted.bam"; |
| 9 | + std::vector<primer> primers = populate_from_file("../data/test_merged.bed"); |
| 10 | + std::string region_; |
| 11 | + samFile *in = hts_open(bam.c_str(), "r"); |
| 12 | + hts_idx_t *idx = sam_index_load(in, bam.c_str()); |
| 13 | + if (idx == NULL) { |
| 14 | + if (sam_index_build2(bam.c_str(), 0, 0) < 0) { |
| 15 | + std::cerr << ("Unable to open BAM/SAM index.") << std::endl; |
| 16 | + return -1; |
| 17 | + } else { |
| 18 | + idx = sam_index_load(in, bam.c_str()); |
| 19 | + if (idx == NULL) { |
| 20 | + std::cerr << "Unable to create BAM/SAM index." << std::endl; |
| 21 | + return -1; |
| 22 | + } |
| 23 | + } |
| 24 | + } |
| 25 | + bam_hdr_t *header = sam_hdr_read(in); |
| 26 | + region_.assign(header->target_name[0]); |
| 27 | + std::string temp(header->text); |
| 28 | + hts_itr_t *iter = NULL; |
| 29 | + iter = sam_itr_querys(idx, header, region_.c_str()); |
| 30 | + bam1_t *aln = bam_init1(); |
| 31 | + cigar_ t; |
| 32 | + uint32_t *cigar; |
| 33 | + int primer_ctr = 0; |
| 34 | + int forward_primer_indices[] = {1, 1, 5, 5, 6}; |
| 35 | + int rev_primer_indices[] = {3, -1, -1, -1, 4}; |
| 36 | + uint8_t cigar_flag[5][16] = { |
| 37 | + {BAM_CSOFT_CLIP, BAM_CSOFT_CLIP, BAM_CMATCH, BAM_CDEL, BAM_CMATCH, BAM_CINS, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CDEL, BAM_CMATCH, BAM_CSOFT_CLIP, BAM_CSOFT_CLIP, BAM_CSOFT_CLIP}, // 3S15M5D2M1D14M1I56M1P6M3P93M2D4M2I8M |
| 38 | + {BAM_CSOFT_CLIP, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CPAD, BAM_CMATCH}, |
| 39 | + {BAM_CSOFT_CLIP, BAM_CSOFT_CLIP, BAM_CSOFT_CLIP, BAM_CSOFT_CLIP, BAM_CSOFT_CLIP, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CSOFT_CLIP}, |
| 40 | + {BAM_CSOFT_CLIP, BAM_CSOFT_CLIP, BAM_CSOFT_CLIP, BAM_CMATCH, BAM_CPAD, BAM_CMATCH}, |
| 41 | + {BAM_CSOFT_CLIP, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CSOFT_CLIP} |
| 42 | + }; |
| 43 | + uint32_t cigar_len[5][16] = { |
| 44 | + {3, 15, 2, 1, 14, 1, 56, 1, 6, 3, 93, 2, 1, 3, 2, 8}, |
| 45 | + {1, 19, 1, 56, 1, 6, 3, 99, 2, 65}, |
| 46 | + {8, 13, 1, 6, 3, 99, 2, 62, 3}, |
| 47 | + {9, 10, 3, 97, 2, 26}, |
| 48 | + {18, 71, 2, 89, 18} |
| 49 | + }; |
| 50 | + uint32_t read_start_pos[5] = { |
| 51 | + 94, 92, 173, 175, 201 |
| 52 | + }; |
| 53 | + uint8_t condense_cigar_flag[5][14] = { |
| 54 | + {BAM_CSOFT_CLIP, BAM_CMATCH, BAM_CDEL, BAM_CMATCH, BAM_CINS, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CDEL, BAM_CMATCH, BAM_CSOFT_CLIP}, |
| 55 | + {BAM_CSOFT_CLIP, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CPAD, BAM_CMATCH}, |
| 56 | + {BAM_CSOFT_CLIP, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CSOFT_CLIP}, |
| 57 | + {BAM_CSOFT_CLIP, BAM_CMATCH, BAM_CPAD, BAM_CMATCH}, |
| 58 | + {BAM_CSOFT_CLIP, BAM_CMATCH, BAM_CPAD, BAM_CMATCH, BAM_CSOFT_CLIP} |
| 59 | + }; |
| 60 | + uint32_t condense_cigar_len[5][13] = { |
| 61 | + {18, 2, 1, 14,1, 56, 1, 6, 3, 93, 2, 1, 13}, |
| 62 | + {1, 19, 1, 56, 1, 6, 3, 99, 2, 65}, |
| 63 | + {31, 99, 2, 62, 3}, |
| 64 | + {22, 97, 2, 26}, |
| 65 | + {18, 71, 2, 89, 18} |
| 66 | + }; |
| 67 | + unsigned int fwd_overlapping_primer_sizes[] = {2, 1, 1, 1, 1}; |
| 68 | + unsigned int rev_overlapping_primer_sizes[] = {2, 0, 0, 0, 1}; |
| 69 | + std::vector<primer> overlapping_primers; |
| 70 | + primer cand_primer; |
| 71 | + while(sam_itr_next(in, iter, aln) >= 0) { |
| 72 | + if((aln->core.flag&BAM_FUNMAP) != 0){ |
| 73 | + continue; |
| 74 | + } |
| 75 | + std::cout << bam_get_qname(aln) << std::endl; |
| 76 | + std::cout << std::endl << "Forward" << std::endl; |
| 77 | + get_overlapping_primers(aln, primers, overlapping_primers, false); |
| 78 | + // Forward primer |
| 79 | + if(overlapping_primers.size() > 0){ |
| 80 | + cand_primer = get_max_end(overlapping_primers); |
| 81 | + t = primer_trim(aln, cand_primer.get_end() + 1, false); |
| 82 | + aln->core.pos += t.start_pos; |
| 83 | + replace_cigar(aln, t.nlength, t.cigar); |
| 84 | + if(overlapping_primers.size() != fwd_overlapping_primer_sizes[primer_ctr]){ |
| 85 | + success = -1; |
| 86 | + std::cout << "Overlapping primer sizes for " << bam_get_qname(aln) <<". Expected: " << fwd_overlapping_primer_sizes[primer_ctr] << ". Got: " << overlapping_primers.size() << std::endl; |
| 87 | + } |
| 88 | + if(cand_primer.get_indice() != forward_primer_indices[primer_ctr]){ |
| 89 | + success = -1; |
| 90 | + std::cout << "Primer indice wrong. Expected: " << forward_primer_indices[primer_ctr] << ". Got: " << cand_primer.get_indice() << std::endl; |
| 91 | + } |
| 92 | + } |
| 93 | + // Reverse primer |
| 94 | + std::cout << std::endl << "Reverse" << std::endl; |
| 95 | + get_overlapping_primers(aln, primers, overlapping_primers, true); |
| 96 | + if(overlapping_primers.size() > 0){ |
| 97 | + cand_primer = get_min_start(overlapping_primers); |
| 98 | + t = primer_trim(aln, cand_primer.get_start() - 1, true); |
| 99 | + replace_cigar(aln, t.nlength, t.cigar); |
| 100 | + if(overlapping_primers.size() != rev_overlapping_primer_sizes[primer_ctr]){ |
| 101 | + success = -1; |
| 102 | + std::cout << "Overlapping primer sizes for " << bam_get_qname(aln) <<". Expected: " << rev_overlapping_primer_sizes[primer_ctr] << ". Got: " << overlapping_primers.size() << std::endl; |
| 103 | + } |
| 104 | + if(cand_primer.get_indice() != rev_primer_indices[primer_ctr]){ |
| 105 | + success = -1; |
| 106 | + std::cout << "Primer indice wrong. Expected: " << rev_primer_indices[primer_ctr] << ". Got: " << cand_primer.get_indice() << std::endl; |
| 107 | + } |
| 108 | + } |
| 109 | + if(aln->core.pos != read_start_pos[primer_ctr]){ |
| 110 | + success = -1; |
| 111 | + std::cout << "Start pos didn't match" << std::endl; |
| 112 | + } |
| 113 | + replace_cigar(aln, t.nlength, t.cigar); |
| 114 | + cigar = bam_get_cigar(aln); |
| 115 | + for (uint i = 0; i < t.nlength; ++i){ |
| 116 | + if(((cigar[i]) & BAM_CIGAR_MASK) != cigar_flag[primer_ctr][i]){ |
| 117 | + success = -1; |
| 118 | + std::cout << "Cigar flag didn't match for " << bam_get_qname(aln) << " ! Expected " << (uint) cigar_flag[primer_ctr][i] << " " << "Got " << ((cigar[i]) & BAM_CIGAR_MASK) << std::endl; |
| 119 | + } |
| 120 | + if((((cigar[i]) >> BAM_CIGAR_SHIFT)) != cigar_len[primer_ctr][i]){ |
| 121 | + success = -1; |
| 122 | + std::cout << "Cigar length didn't match for " << bam_get_qname(aln) << " ! Expected " << (uint) cigar_len[primer_ctr][i] << " " << "Got " << ((cigar[i]) >> BAM_CIGAR_SHIFT) << std::endl; |
| 123 | + } |
| 124 | + } |
| 125 | + // Condense cigar |
| 126 | + std::cout << std::endl << "Condensing cigar ... " << std::endl; |
| 127 | + t = condense_cigar(t.cigar, t.nlength); |
| 128 | + replace_cigar(aln, t.nlength, t.cigar); |
| 129 | + cigar = bam_get_cigar(aln); |
| 130 | + for (uint i = 0; i < t.nlength; ++i){ |
| 131 | + if(((cigar[i]) & BAM_CIGAR_MASK) != condense_cigar_flag[primer_ctr][i]){ |
| 132 | + success = -1; |
| 133 | + std::cout << "Cigar flag didn't match! Expected " << condense_cigar_flag[primer_ctr][i] << " " << "Got " << ((cigar[i]) & BAM_CIGAR_MASK) << std::endl; |
| 134 | + } |
| 135 | + if((((cigar[i]) >> BAM_CIGAR_SHIFT)) != condense_cigar_len[primer_ctr][i]){ |
| 136 | + success = -1; |
| 137 | + std::cout << "Cigar length didn't match after condense! Expected " << condense_cigar_len[primer_ctr][i] << " " << "Got " << ((cigar[i]) >> BAM_CIGAR_SHIFT) << std::endl; |
| 138 | + } |
| 139 | + } |
| 140 | + primer_ctr++; |
| 141 | + } |
| 142 | + // Check if primers found at all |
| 143 | + success = (primer_ctr > 0) ? success : -1; |
| 144 | + return success; |
| 145 | +} |
0 commit comments