kimagure de fo-ku sitemita

moshimokoukouyakyunojoshimanagergaNCBInoSRAwotsukattara.rb

# -*- coding: utf-8 -*-

require "open-uri"
require "json"
require "pp"

###
#  loading bullet
###

organism   = "Homo+sapiens"
study_type = "Transcriptome+Analysis"

###
#   homuhomu
###

def count_samples(namamono, sakusen) # betsuni method teigi suru hitsuyou nakatta kedo maa iiya mitame no mondai desyo konnnano
  # get full study list
  mishima            = "http://trace.ddbj.nig.ac.jp/DRASearch/"
  ground_floor       = "#{mishima}query?organism=#{namamono}&study_type=#{sakusen}&show=100"
  deeper_underground = "#{mishima}query?organism=#{namamono}&study_type=#{sakusen}&show=100&page=2"

  list_study  = URI("#{ground_floor}").read.scan(/href="(study\?acc=.*)" target/).flatten
  list_study2 = URI("#{deeper_underground}").read.scan(/href="(study\?acc=.*)" target/).flatten
  full_list   = list_study + list_study2

  # save full sample list as json format file
  sample_id_full_list = full_list.map { |study_url|
    URI("#{mishima}#{study_url}").read.scan(/href="(sample\?acc=.*)" target/)
  }.flatten

  n = namamono.gsub("+", "_")
  s = sakusen.gsub("+", "_")
  open("./#{n}_#{s}_sample_id.json", "w") {|f| JSON.dump(sample_id_full_list, f)}

  # make hash.. key: sample id, value: sample description
  srs_vs_desc = {}
  full_list.each do |study_url|
    srs_id = URI("#{mishima}#{study_url}").read.scan(/href="(sample\?acc=.*)" target/).flatten

    srs_id.each do |sample_url|
      URI("#{mishima}#{sample_url}").read =~ %r|Description</td><td>(.+)</td>|
      srs_vs_desc[sample_url] = $1 || "no description"
    end
  end

  # save hash (but no need to save him)
  open("./#{n}_#{s}_srs_vs_desc.json", "w") {|f| JSON.dump(srs_vs_desc, f)}

  # reverse hash.. key: sample description, value: sample id (array) 
  desc_vs_srs = {}
  srs_vs_desc.each_pair do |sample_url, desc|
    desc_vs_srs[desc] ||= []
    desc_vs_srs[desc].push(sample_url)
  end

  # count the number of sample ids which have same sample description
  desc_vs_numofsrs = {}
  desc_vs_srs.each_pair do |desc, sample_url|
    numofsrs = sample_url.length
    desc_vs_numofsrs[desc] = numofsrs
  end

  # sort hash by the number of sample ids
  sorted_list = desc_vs_numofsrs.to_a.sort {|a, b|
    (b[1] <=> a[1]) * 2 + (a[0] <=> b[0])
  }

  # fumu~n
  sorted_list
end

if __FILE__ == $0
  # kokode nanika tokubetsu na kotowo suru yotei datta to iu kotonisiteokou
  pp count_samples(organism,study_type)
end