picard_cpp/src/sam/utils/read_ends.h

/*
Description: read ends结构体主要用来标记冗余，包含一些序列的测序过程中的物理信息等

Copyright : All right reserved by ICT

Author : Zhang Zhonghai
Date : 2023/11/3
*/

#ifndef READ_ENDS_H_
#define READ_ENDS_H_

#include <stdint.h>

/**
 * Small interface that provides access to the physical location information about a cluster.
 * All values should be defaulted to -1 if unavailable.  ReadGroup and Tile should only allow
 * non-zero positive integers, x and y coordinates may be negative.
 */
struct PhysicalLocation
{
    /**
     * Small class that provides access to the physical location information about a cluster.
     * All values should be defaulted to -1 if unavailable.  Tile should only allow
     * non-zero positive integers, x and y coordinates must be non-negative.
     * This is different from PhysicalLocationShort in that the x and y positions are ints, not shorts
     * thus, they do not overflow within a HiSeqX tile.
     */
    int16_t tile = -1;
    int32_t x = -1;
    int32_t y = -1;
};

/* 包含了所有read ends信息，如picard里边的 ReadEndsForMarkDuplicates*/
struct ReadEnds : PhysicalLocation
{
    /* ReadEnds中的成员变量 */
    /** Little struct-like class to hold read pair (and fragment) end data for duplicate marking. */
    static const int8_t F = 0, R = 1, FF = 2, FR = 3, RR = 4, RF = 5;
    // int16_t libraryId; // 没用，不考虑多样本
    int8_t orientation;
    int32_t read1ReferenceIndex = -1;
    int32_t read1Coordinate = -1;
    int32_t read2ReferenceIndex = -1;
    int32_t read2Coordinate = -1; // This field is overloaded for flow based processing as the end coordinate of read 1. (paired reads not supported)
    /* Additional information used to detect optical dupes */
    // int16_t readGroup = -1; 一般经过比对后的bam文件只有一个read group，normal或者tumor
    /** For optical duplicate detection the orientation matters regard to 1st or 2nd end of a mate */
    int8_t orientationForOpticalDuplicates = -1;
    /** A *transient* flag marking this read end as being an optical duplicate. */
    bool isOpticalDuplicate = false;

    /* ReadEndsForMarkDuplicates中的成员变量 */
    /** Little struct-like class to hold read pair (and fragment) end data for MarkDuplicatesWithMateCigar **/
    int16_t score = 0;
    int64_t read1IndexInFile = -1;
    int64_t read2IndexInFile = -1;
    int64_t duplicateSetSize = -1;

    /* ReadEndsForMarkDuplicatesWithBarcodes中的成员变量 (好像用不到) */
    // int32_t barcode = 0; // primary barcode for this read (and pair)
    // int32_t readOneBarcode = 0; // read one barcode, 0 if not present
    // int32_t readTwoBarcode = 0; // read two barcode, 0 if not present or not paired

    /* zzh增加的成员变量 */
    int64_t posKey = -1; // 根据位置信息生成的关键字 return (int64_t)tid << MAX_CONTIG_LEN_SHIFT | (int64_t)pos;

    /* 根据pairend read的比对方向，来确定整体的比对方向 */
    static int8_t GetOrientationByte(bool read1NegativeStrand, bool read2NegativeStrand)
    {
        if (read1NegativeStrand)
        {
            if (read2NegativeStrand)
                return RR;
            else
                return RF;
        }
        else
        {
            if (read2NegativeStrand)
                return FR;
            else
                return FF;
        }
    }

    /* 比较两个readends是否一样（有个冗余） */
    static bool AreComparableForDuplicates(ReadEnds &lhs, ReadEnds &rhs, bool compareRead2)
    {
        bool areComparable = true;
        areComparable = lhs.read1ReferenceIndex == rhs.read1ReferenceIndex &&
                        lhs.read1Coordinate == rhs.read1Coordinate &&
                        lhs.orientation == rhs.orientation;
        if (areComparable && compareRead2)
        {
            areComparable = lhs.read2ReferenceIndex == rhs.read2ReferenceIndex &&
                            lhs.read2Coordinate == rhs.read2Coordinate;
        }
        return areComparable;
    }

    /* 比对方向是否正向 */
    bool IsForwardStrand()
    {
        return orientation == F;
    }

    /* pairend是否合适的比对上了 */
    bool IsPaired()
    {
        return read2ReferenceIndex != -1;
    }
};

#endif