FastBQSR/src/bqsr/bqsr_entry.cpp

/*
Description:
bam，bam，bam

Copyright : All right reserved by ICT

Author : Zhang Zhonghai
Date : 2023/10/23
*/
#include <htslib/sam.h>
#include <htslib/synced_bcf_reader.h>
#include <htslib/thread_pool.h>
#include <spdlog/spdlog.h>

#include <iomanip>
#include <vector>

#include "bqsr_args.h"
#include "bqsr_funcs.h"
#include "bqsr_pipeline.h"
#include "dup_metrics.h"
#include "fastbqsr_version.h"
#include "read_name_parser.h"
#include "util/profiling.h"
#include "util/utils.h"

#define BAM_BLOCK_SIZE 16L * 1024 * 1024

namespace nsgv {


std::vector<ReadNameParser> gNameParsers;  // read name parser
DuplicationMetrics gMetrics;               //
DupResult gDupRes;
PipelineArg gPipe(&gDupRes);

BQSRArg gBqsrArg;                     //
samFile *gInBamFp;                    // bam
sam_hdr_t *gInBamHeader;              // bam
samFile *gOutBamFp;                   // , sambam
sam_hdr_t *gOutBamHeader;             // header
vector <bcf_srs_t*> gKnownSitesVcfSrs; // known sites vcf srs
};  // namespace nsgv

// 
struct ByteBuf {
    uint8_t *buf = nullptr;
    int size = 0;      // 
    int capacity = 0;  // 
};

/*
 * 
 */
static string getFileExtension(const string &filename) {
    auto last_dot = filename.find_last_of('.');
    if (last_dot == string::npos) {
        return "";
    }
    return filename.substr(last_dot + 1);
}

// 串行bqsr
int SerialBQSR() {
    int round = 0;
    BamBufType inBamBuf(nsgv::gBqsrArg.DUPLEX_IO);
    inBamBuf.Init(nsgv::gInBamFp, nsgv::gInBamHeader, nsgv::gBqsrArg.MAX_MEM);
    int64_t readNumSum = 0;
    while (1) {
        ++ round;
        size_t readNum = 0;
        if (inBamBuf.ReadStat() >= 0)
            readNum = inBamBuf.ReadBam();
        if (readNum < 1) {
            break;
        }
        spdlog::info("{} reads processed in {} round", readNum, round);

        auto bams = inBamBuf.GetBamArr();
        spdlog::info("region: {} - {}", bams[0]->softclip_start(), bams.back()->softclip_end());
        // 1. 获取bams数组覆盖的region范围

        // 2. 开辟一个uint32_t的数组作为bitmap（如果上一轮的不够就重开），用来表示region的每个位点是否有known sites覆盖（每轮使用前需清零）

        // 3. 读取在region范围内的所有known sites，并为对应的bitmap设定0 or 1 (作为skip标识)

        // 4. 遍历bams数组中的每一条记录并进行处理

        readNumSum += readNum;
        inBamBuf.ClearAll();     // 
    }
    spdlog::info("read count: {}", readNumSum);

    return 0;
}

// entrance of mark duplicates
int BaseRecalibrator() {
    
    PROF_START(whole_process);
    /* bam */
    nsgv::gInBamFp = sam_open_format(nsgv::gBqsrArg.INPUT_FILE.c_str(), "r", nullptr);
    if (!nsgv::gInBamFp) {
        spdlog::error("[{}] load sam/bam file failed.\n", __func__);
        return -1;
    }
    hts_set_opt(nsgv::gInBamFp, HTS_OPT_BLOCK_SIZE, BAM_BLOCK_SIZE);
    nsgv::gInBamHeader = sam_hdr_read(nsgv::gInBamFp);  // header
    
    // (libraryId)
    nsgv::gMetrics.LIBRARY = sam_hdr_line_name(nsgv::gInBamHeader, "RG", 0);

    /* 并行读取bam数据  */
    htsThreadPool htsPoolRead = {NULL, 0};   // ，
    htsThreadPool htsPoolWrite = {NULL, 0};  // 
    htsPoolRead.pool = hts_tpool_init(nsgv::gBqsrArg.NUM_THREADS);
    htsPoolWrite.pool = hts_tpool_init(nsgv::gBqsrArg.NUM_THREADS);
    if (!htsPoolRead.pool || !htsPoolWrite.pool) {
        spdlog::error("[{}] failed to set up thread pool", __LINE__);
        sam_close(nsgv::gInBamFp);
        return -1;
    }
    hts_set_opt(nsgv::gInBamFp, HTS_OPT_THREAD_POOL, &htsPoolRead);

    return SerialBQSR();

//    // 读取known sites vcfs
//    for (const auto& ks : nsgv::gBqsrArg.KNOWN_SITES_VCFS) {
//        spdlog::info("  {}", ks);
//        bcf_srs_t* srs = bcf_sr_init();
//        if (!bcf_sr_add_reader(srs, ks.c_str()))
//            error("Failed to read from %s: %s\n", !strcmp("-", ks.c_str()) ? "standard input" : ks.c_str(), bcf_sr_strerror(srs->errnum));
//        nsgv::gKnownSitesVcfSrs.push_back(srs);
//
//        while (bcf_sr_next_line(srs)) {
//            bcf1_t* line = srs->readers[0].buffer[0];
//            cout << line->pos << '\t' << line->rlen << '\t' << line->n_allele << '\t' << line->n_info << endl;
//        }
//    }
//
//    /* 先实现串行的bqsr-phase-1 */
    




    sam_close(nsgv::gInBamFp);

    PROF_END(gprof[GP_whole_process], whole_process);

    return 0;
}