#! /usr/bin/ruby
#
# Author:: Pjotr Prins
# Copyright:: August 2010
# License:: Ruby License
#
# Copyright (C) 2010,2011 Pjotr Prins <pjotr.prins@thebird.nl>

USAGE = <<EOM

  Fetch and assemble GFF3 types (e.g. ORF, mRNA, CDS) + print in FASTA format. 

    gff3-fetch [--low-mem] [--validate] type [filename.fa] filename.gff3

  Where (NYI == Not Yet Implemented):

    --translate     : output as amino acid sequence 
    --validate      : validate GFF3 file by translating
    --fix           : check 3-frame translation and fix, if possible 
    --fix-wormbase  : fix 3-frame translation on ORFs named 'gene1'
    --no-assemble   : output each record as a sequence -- NYI
    --add-phase     : output records using phase (useful w. no-assemble CDS to AA) --NYI
  
  type is any valid type in the GFF3 definition. For example:

    mRNA            : assemble mRNA
    CDS             : assemble CDS 
    exon            : list all exons
    gene/ORF        : list gene ORFs -- NYI
    other           : use any type from GFF3 definition, e.g. 'Terminate' -- NYI

  and the following performance options:

    --cache full    : load all in RAM (fast)
    --cache none    : do not load anything in memory (slow)
    --low-mem       : use LRU cache (limit RAM use, fast) -- NYI
    --max-cpus num  : use num threads -- NYI
    --emboss        : use EMBOSS translation (fast) -- NYI

  Multiple GFF3 files can be used. With external FASTA files, always the last
  one before the GFF3 filename is matched.

  Note that above switches are only partially implemented at this stage. Full
  feature support is projected Feb. 2011.

  Examples:

    Assemble mRNA and CDS information from test.gff3 (which includes sequence information)

      gff3-fetch mRNA test/data/gff/test.gff3
      gff3-fetch CDS test/data/gff/test.gff3

    Find CDS records from external FASTA file, adding phase and translate to protein sequence

      gff3-fetch --no-assemble --add-phase --translate CDS test/data/gff/MhA1_Contig1133.fa test/data/gff/MhA1_Contig1133.gff3

    Find mRNA from external FASTA file, without loading everything in RAM

      gff3-fetch --cache none mRNA test/data/gff/test-ext-fasta.fa test/data/gff/test-ext-fasta.gff3   
      gff3-fetch --cache none mRNA test/data/gff/test-ext-fasta.fa test/data/gff/test-ext-fasta.gff3   

    Validate GFF3 file using EMBOSS translation and validation

      gff3-fetch --cache none --validate --emboss mRNA test/data/gff/test-ext-fasta.fa test/data/gff/test-ext-fasta.gff3   

    Find GENEID predicted terminal exons

      gff3-fetch terminal chromosome1.fa geneid.gff3

== Performance

time gff3-fetch cds m_hapla.WS217.dna.fa m_hapla.WS217.gff3 > test.fa

  Cache      real     user     sys
  ----------------------------------------------------
  full       12m41s   12m28s   0m09s (0.8.0 Jan. 2011)
  none      504m39s  477m49s  26m50s (0.8.0 Jan. 2011)
  ----------------------------------------------------

where

   52M m_hapla.WS217.dna.fa
  456M m_hapla.WS217.gff3

ruby 1.9.2p136 (2010-12-25 revision 30365) [x86_64-linux]
on an 8 CPU, 2.6 GHz (6MB cache), 16 GB RAM machine. 

== Cite

  If you use this software, please cite 
  
    http://dx.doi.org/10.1093/bioinformatics/btq475

== Copyright

Copyright (C) 2010,2011 Pjotr Prins <pjotr.prins@thebird.nl> 

EOM

SEP = File::SEPARATOR
rootpath = File.dirname(File.dirname(__FILE__))+SEP
$: << rootpath+'lib'
GFF3_VERSION = File.new(rootpath+'VERSION').read.chomp

require 'bio-gff3'
require 'optparse'
require 'ostruct'
require 'bio/output/gfffastawriter'

$stderr.print "BioRuby GFF3 Plugin "+GFF3_VERSION+" Copyright (C) 2010,2011 Pjotr Prins <pjotr.prins@thebird.nl>\n\n"

if ARGV.size == 0
  print USAGE
end

options = OpenStruct.new()
opts = OptionParser.new() { |opts|
  opts.on_tail("-h", "--help", "Show help and examples") {
    print(opts)
    print USAGE
    exit()
  }

  opts.on("--cache [none,full]", String, "Caching (default full)") do |cache|
    case cache.downcase
      when 'none'
        options.cache = :cache_none
      # when 'lru'
      #   options.cache = :cache_lru
      when 'full'
        options.cache = :cache_full
      else
        raise 'Unknown --cache option'
    end
  end

  opts.on("--translate", "output as amino acid sequence") do |b|
    options.translate = b
  end

  opts.on("--validate", "validate GFF3 file by translating") do |v|
    options.validate = v
    $stop_on_error = true  # replace global in near future
  end

  opts.on("--fix", "Fix frame errors in the GFF3 definition") do |v|
    options.fix = true
  end

  opts.on("--fix-wormbase", "Wormbase fix gene1 frame error") do |v|
    options.fix_wormbase = true
  end

  # opts.on("-q", "--quiet", "Run quietly") do |q|
  #   options.quiet = q
  # end

  # opts.on("-v", "--[no-]verbose", "Run verbosely") do |v|
  #   options.verbose = v
  # end

  # opts.on("-t", "--[no-]trace", "Debug") do |t|
  #   options.trace = t
  # end
}
opts.parse!(ARGV)

gfftype = ARGV.shift

raise "Unknown GFF type '#{gfftype}'" if gfftype !~ /mrna|cds|exon/i

fastafn = nil

ARGV.each do | fn |
  if File.extname(fn) =~ /fa|fas|fasta$/i
    raise "Only one fasta file allowed per GFF3" if fastafn != nil
    fastafn = fn
    next
  end

  opts = {}
  opts[:validate] = options.validate
  opts[:cache_components] = options.cache
  opts[:cache_records]    = options.cache
  opts[:fasta_filename] = fastafn if fastafn
  opts[:fix_wormbase] = options.fix_wormbase
  opts[:fix] = options.fix
  gffdb = Bio::GFFbrowser::GFFdb.new(fn,opts)
  gff = gffdb.assembler
  writer = Bio::GFFbrowser::FastaWriter.new(options.translate, options.validate)
  case gfftype.downcase
    when 'mrna' 
          gff.each_mRNA_seq do | id, seq |
            writer.put(id,seq)
          end
    when 'exon'
          gff.each_exon_seq do | id, seq |
            writer.put(id,seq)
          end
    when 'cds'
          gff.each_CDS_seq do | id, seq |
            writer.put(id,seq)
          end
    else
      raise "Unknown action <#{gfftype}>"
  end
  fastafn = nil
end