gtf.sublime-syntax

%YAML 1.2
---
# Flagstat syntax highlighting file
# Maintainer: bioSyntax.org
# Version: v0.1
name: gtf
file_extensions: [gtf]
scope: source.gtf
# Specification: http://mblab.wustl.edu/GTF22.html

contexts:
  main:
    - match: '^[\S]*\t'
      scope: chr.gtf
      push: source

  source:
    - match: '[\S]*'
      scope: variable.parameter.gtf
    - match: \t
      push: feature
    - match: $
      pop: true

  feature:
    - match: 'exon'
      scope: exonLevel.gtf
    - match: 'transcript'
      scope: transcriptLevel.gtf
    - match: 'CDS'
      scope: cds.gtf
    - match: 'start_codon'
      scope: start_codon.gtf
    - match: 'stop_codon'
      scope: stop_codon.gtf
    - match: \t
      push: start
    - match: $
      pop: true

  start:
    - match: ([0-9|\.](?=\d{2}\t))
      scope: chrStart2.bed
    - match: ([0-9|\.](?=\d{5}\t))
      scope: chrStart2.bed
    - match: ([0-9|\.](?=\d{8}\t))
      scope: chrStart2.bed
    - match: ([0-9|\.])
      scope: chrStart.bed
    - match: \t
      push: end
    - match: $
      pop: true

  end:
    - match: ([0-9|\.](?=\d{2}\t))
      scope: chrStart2.bed
    - match: ([0-9|\.](?=\d{5}\t))
      scope: chrStart2.bed
    - match: ([0-9|\.](?=\d{8}\t))
      scope: chrStart2.bed
    - match: ([0-9|\.])
      scope: chrEnd.bed
    - match: \t
      push: score
    - match: $
      pop: true

  score:
    - match: \.\b
      scope: comment.gtf
    - match: ([1-9])\b
      scope: grad1.gtf  
    - match: ([1-9][0-9])\b
      scope: grad1.gtf
    - match: 1[0-9][0-9]\b
      scope: grad2.gtf
    - match: 2[0-9][0-9]\b
      scope: grad3.gtf
    - match: 3[0-9][0-9]\b
      scope: grad4.gtf
    - match: 4[0-9][0-9]\b
      scope: grad5.gtf
    - match: 5[0-9][0-9]\b
      scope: grad6.gtf
    - match: 6[0-9][0-9]\b
      scope: grad7.gtf
    - match: 7[0-9][0-9]\b
      scope: grad8.gtf
    - match: 8[0-9][0-9]\b
      scope: grad9.gtf
    - match: 9[0-9][0-9]\b
      scope: grad10.gtf  
    - match: ([0-9][0-9][0-9][0-9])\b
      scope: grad10.gtf                
    - match: \t
      push: strand
    - match: $
      pop: true

  strand:
    - match: \+
      scope: strandPlus.gtf
    - match: \-
      scope: strandMinus.gtf
    - match: \.
      scope: strandNone.gtf
    - match: \t
      push: frame
    - match: $
      pop: true

  frame:
    - match: ([0-2])\b
      scope: numeric.gtf
    - match: \.
      scope: comment.gtf
    - match: \t
      push: attribute
    - match: $
      pop: true

  attribute:
    - match: \b[^\s;]+(?=\s"[\w|\W|\_|\.]*")
      scope: gradbw5.attributeName.gtf
    - match: (?="[\w|\d|\.]+";)"[\d|\.]+"
      scope: numeric.attributeNumbers.gtf
    - match: (?="[\w|\d|\.|\-|\_|\:]+";)"[\w|\d|\.|\-|\_|\:]+"
      scope: grad4.attributeString.gtf
    - match: \b[^\s;]+(?=\s[\w|\W|\_|\.])
      scope: gradbw5.attributeName.gtf
    - match: (?=\d+;)\d+
      scope: numeric.attributeNumbers.gtf
    - match: \t
      push: attribute
    - match: $
      pop: true