gistfile1.rb

#!/users/p-solid/local/bin/ruby19
#
# If you want to run against multiple data sets:
# ruby -e 'File.open(Dir["list*"].join).each_line{|l| puts "./create_splits.rb {File.basename(l.chomp)} #{l.chomp}"}'
#
require 'fileutils'

def run_lsf(str)
  bs = "bsub -q solid -J GBM_drio_splits '#{str}'"
  puts bs
  system bs
end

def run(str)
  puts str
  system str
end

split_size = "25000000"
run_name   = ARGV[0]
csfasta    = ARGV[1]

if ARGV.size != 2
  puts "Usage: #{__FILE__} <run_name> <csfasta>"
  exit 1
end

raise "./splits not found" unless File.exists?("./splits")
Dir.chdir("./splits")

raise "#{run_name} dir already exists" if File.exists?(run_name)
FileUtils.mkdir run_name 
Dir.chdir(run_name)

run_lsf "cat #{csfasta} | grep -v '^#' | split -a4 -d -l #{split_size}"