gentoo-overlay/sci-biology/goby-cpp/files/Reads.proto

package goby;

option java_package = "edu.cornell.med.icb.goby.reads";
option optimize_for = SPEED;

message ReadCollection {
     repeated ReadEntry reads = 1;
}

message ReadEntry {
  /*
    Index of a read.
  */
  required uint32 read_index = 1;
   /*
    Index of the barcode, if any.
  */
  optional uint32 barcode_index = 10;
  /*
     Read identifier/name may be present.
  */
  optional string read_identifier = 23;
  /*
     Additional description about the read (from Fasta/Q format).
   */
  optional string description = 22;
  /*
    Length of the sequence.
   */
  required uint32 read_length = 2;
  /*
    Sequence, encoded as ascii characters stored in single bytes.
   */
  optional bytes sequence = 3;
  /*
   The second sequence in a pair. Stored the same way as the sequence attribute.
  */
  optional bytes sequence_pair = 5;
  /*
    Length of the second sequence in a pair.
  */
  optional uint32 read_length_pair = 6;
  /*
    Quality scores in Phred units, stored as single bytes (0-255).
  */
  optional bytes quality_scores = 4;
  /*
    Quality scores for the second sequence in a pair. Stored as the 'qualityScores' attribute.
   */
  optional bytes quality_scores_pair = 7;
  /*
    Compressed stream of data. The first byte indicates the compression/decompression method (codec). The remaining bytes are
    content compressed with the codec.
  */
  optional bytes compressed_data = 8;
  /*
     Stores meta-data about the reads. Typically meta-data is stored in the very first read of a
     read collection, with the understanding that the meta-data applies to all the reads in the
     collection. Meta-data can be used to store information about when the sample was sequenced,
     or other information of interest. The key-value pair format is sufficiently flexible to
     accomodate a variety of needs. The following keys are pre-defined. Please use pre-defined
     keys so that automated tools can use metadata in relatively standard way. Please note that
     some keys provide a format for the value. This format should also be followed to garantee
     that meta data can be used computationally in fully automatic manner.

     key="sequencing-run-start-date" value="MM/DD/YYYY" Used to record when the sequencing run
     was initiated on the instrument. Can be used to detect batch effect in a large set of samples.
     key="platform" value="<free-text>". Value is free text, but the following terms are pre-defined.
      Illumina GaIIx
      Illumina HiSeq 1000
      Illumina HiSeq 2000
      Helicos Heliscope
      LifeTech 5500 SOLiD
      LifeTech 5500xl SOLiD
      Roche 454 GS FLX Ti

      key="organism" value="species name"
      Since Goby 1.9.1
  */
  repeated MetaData meta_data = 25;

}
/*
 A message to store a key/value pair and represent metadata about reads.
 Since Goby 1.9.1
 */
message MetaData {
 /*
   Provides the key. See examples in the documentation of meta_data for ReadEntry.
 */
 required string key=1;
 /*
   Describes the value associated with the key. See examples in the documentation of meta_data for ReadEntry.
 */
 required string value=2;
}