bed reader

gistfile1.d

/**
 * The module csv is a set of function to parse many format using a delimitter as csv file
 * Supported format:
 *  - .mat matrix file
 *  - .bed UCSC file
 * For parse a .csv file use std.csv
 */
module dscience.parser.csv;

import std.conv : to;
import std.csv;
import std.file;
import std.array : array, map, split, empty;
import std.algorithm : filter;
import std.string;

struct Bed3{
    string    chrom;        // 0
    size_t    chromStart;   // 1
    size_t    chromEnd;     // 2
    string    name;         // 3
}

struct Bed4{
    string    chrom;        // 0
    size_t    chromStart;   // 1
    size_t    chromEnd;     // 2
    string    name;         // 3
    size_t    score;        // 4
}

struct Bed5{
    string    chrom;        // 0
    size_t    chromStart;   // 1
    size_t    chromEnd;     // 2
    string    name;         // 3
    size_t    score;        // 4
    char      strand;       // 5
}

struct Bed6{
    string    chrom;        // 0
    size_t    chromStart;   // 1
    size_t    chromEnd;     // 2
    string    name;         // 3
    size_t    score;        // 4
    char      strand;       // 5
    size_t    thickStart;   // 6
}

struct Bed7{
    string    chrom;        // 0
    size_t    chromStart;   // 1
    size_t    chromEnd;     // 2
    string    name;         // 3
    size_t    score;        // 4
    char      strand;       // 5
    size_t    thickStart;   // 6
    size_t    thickEnd;     // 7
}

struct Bed8{
    string    chrom;        // 0
    size_t    chromStart;   // 1
    size_t    chromEnd;     // 2
    string    name;         // 3
    size_t    score;        // 4
    char      strand;       // 5
    size_t    thickStart;   // 6
    size_t    thickEnd;     // 7
    size_t[3] itemRgb;      // 8
}

struct Bed9{
    string    chrom;        // 0
    size_t    chromStart;   // 1
    size_t    chromEnd;     // 2
    string    name;         // 3
    size_t    score;        // 4
    char      strand;       // 5
    size_t    thickStart;   // 6
    size_t    thickEnd;     // 7
    size_t[3] itemRgb;      // 8
    size_t    blockCount;   // 9
}

struct Bed10{
    string    chrom;        // 0
    size_t    chromStart;   // 1
    size_t    chromEnd;     // 2
    string    name;         // 3
    size_t    score;        // 4
    char      strand;       // 5
    size_t    thickStart;   // 6
    size_t    thickEnd;     // 7
    size_t[3] itemRgb;      // 8
    size_t    blockCount;   // 9
    size_t    blockSizes;   // 10
}

struct Bed11{
    string    chrom;        // 0
    size_t    chromStart;   // 1
    size_t    chromEnd;     // 2
    string    name;         // 3
    size_t    score;        // 4
    char      strand;       // 5
    size_t    thickStart;   // 6
    size_t    thickEnd;     // 7
    size_t[3] itemRgb;      // 8
    size_t    blockCount;   // 9
    size_t    blockSizes;   // 10
    size_t    blockStarts;  // 11
}


struct BedMetadata{
    string name;
    string description;
    size_t visibility;
    string itemRgb;
    size_t browserStart;
    size_t browserEnd;
    string chromosome;
    string hide;

    string toString(){
        string result = "";
        if(  chromosome != "" && browserStart != 0 && browserEnd != 0 )
            result ~= "browser position %s:%d-%d\n".format( chromosome, browserStart, browserEnd );
        if( hide != "" )
            result ~= "browser hide %s\n".format( hide );
        if( name != "" && description != "" && visibility != 0 )
            result ~= "track name=%s description=%s visibility=%d\n".format( name, description, visibility );
        if( itemRgb != "" )
            result ~= "itemRgb=\"%s\"".format( itemRgb );
        return result;
    }
}

BedMetadata bedReader( T=Bed3)( string filePath, char delimiter='\t', CsvReader delegate(char) dg ){
    if( !filePath.exist )
        throw new  FileException( "File %s is do not exist".format(filePath) );
    else if( !filePath.isFile )
        throw new  FileException( "File %s is not a file".format(filePath) );

    File        bedFile = File( filePath, "r" );
    BedMetadata metadata;

    foreach( char[] line, bedFile.byLine() ){
        if( line[0] == '#' )    // comment
            continue;
        else if( line == "" )   // empty line
            continue;
        else if( line[0..16] == "browser position" ){
            // todo
        }
        else if( line[0..12] == "browser hide" ){
            // todo
        }
        else{                   // data in csv format
            CsvReader records = csvReader!T(line, delimiter);
            dg(records);
        }
    }

    return metadata;
}