sam.sublime-syntax

%YAML 1.2
---
# SAM syntax highlighting file
# Maintainer: bioSyntax.org
# Version: v0.1
# See http://www.sublimetext.com/docs/3/syntax.html
name: sam
file_extensions:
  - sam
scope: source.sam
contexts:
  main:
    - match: '^@HD'
      scope: header.sam
      push: header_entry_hd

    - match: '^@SQ'
      scope: header.sam
      push: header_entry_sq

    - match: '^@RG'
      scope: header.sam
      push: header_entry_rg

    - match: '^@PG'
      scope: header.sam
      push: header_entry_pg

    - match: '^@CO'
      scope: header.sam
      push: header_entry_co

    # entering sequence alignment section
    # column 1 of sequence alignment: QNAME
    - match: '^[\S]*\t'
      scope: keyword.sam
      push: flag_column


  # header entries

  header_entry_hd:
    - meta_scope: string.sam
    - match: '(?<=[A-Z]{2}:)"'
      scope: invalid.sam
      push: header_entry_quote
    - match: '(?<=VN:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: numeric.sam

    # Sort Order for sam file
    # Distinguishes sorted from unsorted files
    - match: '(?<=SO:)(coordinate|queryname(?=\s+[A-Z]{2}:)|.+$)'
      scope: bon.sam
    - match: '(?<=SO:)(unsorted|unknown(?=\s+[A-Z]{2}:)|.+$)'
      scope: mal.sam
    - match: '(?<=SO:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam

    - match: '(?<=GO:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '$'
      pop: true

  header_entry_sq:
    - meta_scope: string.sam
    - match: '(?<=[A-Z]{2}:)"'
      scope: invalid.sam
      push: header_entry_quote
    - match: '(?<=SN:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: chr.sam
    - match: '(?<=LN:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: chrStart.sam
    - match: '(?<=AH:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=AN:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=AS:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=M5:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=SP:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=UR:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: $
      pop: true

  header_entry_rg:
    - meta_scope: string.sam
    - match: '(?<=[A-Z]{2}:)"'
      scope: invalid.sam
      push: header_entry_quote
    - match: '(?<=ID:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: library.sam
    - match: '(?<=LB:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: library.sam
    - match: '(?<=CN:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=DS:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=DT:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=FO:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=KS:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=PG:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: platform.sam
    - match: '(?<=PI:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=PL:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: platform.sam
    - match: '(?<=PM:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: keyword.sam
    - match: '(?<=PU:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: platform.sam
    - match: '(?<=SM:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: library.sam
    - match: $
      pop: true

  header_entry_pg:
    - meta_scope: string.sam
    - match: '(?<=[A-Z]{2}:)"'
      scope: variable.parameter.sam
      push: header_entry_quote
    - match: '(?<=ID:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: software.sam
    - match: '(?<=PN:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: software.sam
    - match: '(?<=CL:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: command.sam
    - match: '(?<=PP:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: software.sam
    - match: '(?<=DS:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: string.sam
    - match: '(?<=VN:)(.+?(?=\s+[A-Z]{2}:)|.+$)'
      scope: numeric.sam
    - match: $
      pop: true

  header_entry_co:
    - meta_scope: string.sam
    - match: '(?<=[A-Z]{2}:)"'
      scope: comment.am
      push: header_entry_quote
    - match: $
      pop: true

  header_entry_quote:
    - meta_scope: variable.parameter.sam
    - match: '"'
      pop: true


  # sequence alignment columns

  # column 2: FLAG
  flag_column:
    # Read Flag
    # If flag > 512
    #   read is QC-fail | PCRdup | 2' alignment
    - match: '[0-9]{1,2}(?=\t)'
      scope: numeric.sam
    - match: '[1-4][0-9]{1,2}(?=\t)'
      scope: numeric.sam
    - match: '[5][0][0-9](?=\t)'
      scope: numeric.sam
    - match: '[5][1][01](?=\t)'
      scope: numeric.sam

    - match: '[5][1][01]'
      scope: ntC.sam
    - match: '[\S]*'
      scope: ntC.sam

    - match: \t
      push: rname_column
    - match: $
      pop: true

  # column 3: RNAME: usually a chromosome id
  rname_column:
    - match: '[\S]*'
      scope: chr.sam
    - match: \t
      push: pos_column
    - match: $
      pop: true

  # column 4: POS: 1-based leftmost mapping POSition of the first CIGAR operation that “consumes” a reference base
  pos_column:
    - match: '[\S]*'
      scope: chrStart.sam
    - match: \t
      push: mapq_column
    - match: $
      pop: true

  # column 5: MAPQ: MAPping Quality. 
  mapq_column:
    - match: '(255|[0-9])(?=\t)'
      scope: gradbw1
    - match: '1[0-9](?=\t)'
      scope: gradbw3
    - match: '2[0-9](?=\t)'
      scope: gradbw5
    - match: '3[0-9](?=\t)'
      scope: gradbw7
    - match: '4[0-9](?=\t)'
      scope: gradbw9
    - match: '[\S]*'
      scope: gradbw9

    - match: \t
      push: cigar_column
    - match: $
      pop: true


  # column 6: CIGAR: Alignment details
  cigar_column:
      # CIGAR Strings
      # M: Alignment match
      # I: Insertion
      # D: Deletion
      # N: Skipped region (intron)
      # S: Soft Clip
      # H: Hard Clip
      # P: Padding
      # =: Sequence Match
      # X: Sequence Mismatch
      # *: String Absent

    - match: '\d+[NSHP]'
      scope: cigClip.sam
    - match: '\d+[M=]'
      scope: cigMat.sam
    - match: '\d+[X]'
      scope: cigMis.sam
    - match: '\d+[D]'
      scope: cigDel.sam
    - match: '\d+[I]'
      scope: cigIns.sam
    - match: '[\S]*'
      scope: comment.sam
    - match: \t
      push: rnext_column
    - match: $
      pop: true

  # column 7: RNEXT Reference sequence name of the primary alignment of the NEXT read in the template.
  rnext_column:
    - match: "="
      scope: chr.sam
      push: pnext_column_gateway
    - match: "[0-9]*(?=\t)"
      scope: chrStart.sam
      push: pnext_column_gateway
    - match: "\\*"
      scope: keyword.sam
      push: pnext_alt_seq_column_gateway
    - match: '[\S]*(?=\t)'
      scope: keyword.Start
      push: pnext_alt_seq_column_gateway
    - match: $
      pop: true

  # column 8: PNEXT: Position of the primary alignment of the NEXT read in the template. 
  
  pnext_column_gateway:
    - match: \t
      push: pnext_column
    - match: $
      pop: true

  pnext_column:
    # - meta_scope: invalid.sam
    - match: '[\S]*'
      scope: chrStart.sam
    - match: \t
      push: tlen_column
    - match: $
      pop: true

  pnext_alt_seq_column_gateway:
    - match: \t
      push: pnext_alt_seq_column
    - match: $
      pop: true

  pnext_alt_seq_column:
    # - meta_scope: invalid.sam
    - match: '[\S]*'
      scope: keyword.Start
    - match: \t
      push: tlen_column
    - match: $
      pop: true

  # column 9: TLEN:  signed  observed  Template  LENgth.   
  tlen_column:
    - match: '[\S]*'
      scope: comment.sam
    - match: \t
      push: seq_column
    - match: $
      pop: true

  # column 10: SEQ: segment SEQuence.
  seq_column:
    - match: "A"
      scope: ntA.sam
    - match: "C"
      scope: ntC.sam
    - match: "G"
      scope: ntG.sam
    - match: "T"
      scope: ntT.sam
    - match: "N"
      scope: ntN.sam
    - match: \t
      push: qual_column
    - match: $
      pop: true

  # column 11: QUAL: ASCII of base QUALity plus 33 (same as the
  # quality string in the Sanger FASTQ format).
  # Complete Phred Scale
  #!"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~

  # Illumina 1.8 Scale (IMPLEMENTED)
  # 0.2.......................26...31.......41     
  # !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKL
  # |   |   |   |   |   |   |   |   |   |   | ...

  qual_column:
    - match: "[\\!\\#\\$]"
      scope: gradbw0
    - match: "[%\\&\\'\\(]"
      scope: gradbw1
    - match: "[\\)\\*\\+\\,]"
      scope: gradbw2
    - match: "[\\-\\.\\/0]"
      scope: gradbw3
    - match: "[1-4]"
      scope: gradbw4
    - match: "[5-8]"
      scope: gradbw5
    - match: "[9\\:\\;\\<]"
      scope: gradbw6
    - match: "[\\=\\>\\?\\@]"
      scope: gradbw7
    - match: "[ABCD]"
      scope: gradbw8
    - match: "[EFGHIJ]"
      scope: gradbw9
    - match: "[K-Z\\[\\\\]"
      scope: gradbw10
    - match: "[\\]\\^\\_\\`a-f]"
      scope: gradbw10
    - match: "[g-p]"
      scope: gradbw10
    - match: "[q-z]"
      scope: gradbw10
    - match: "[\\{\\|\\}\\~]"
      scope: gradbw10
    - match: \t
      push: data_tag_entry
    - match: $
      pop: true

  data_tag_entry: 
    - meta_scope: string.sam
      # Data Classes: numeric / string / class
      # A Printable
      # i signed integer
      # f floating point
      # Z string with spaces
      # H byte array hex
      # B Integer/Numeric Array
    - match: "[A-Z]{2}(?=:)"
      scope: string.sam
    - match: "(?<=:)[AifZHB]:"
      scope: comment.sam

    - match: "(?<=[ifH]:)(.+?(?=\t)|.+$)"
      scope: numeric.sam
    - match: "(?<=[AZB]:)(.+?(?=\t)|.+$)"
      scope: numeric2.sam

    - match: "$"
      pop: true