my @objfuns = ('de', 'cd');
my @datasets = ('crp0', 'adh', 'Klf1-200-100', 'Klf1-200');
my $n_basic_dsets = @datasets;
if (&is_comprehensive()) {
  push(@datasets, 'INO_up800', 'farntrans5', 'lipocalin', 'Puf3p', 'dna_with_covalent');
}
my %params = (
  'crp0' => ['-dna', '-usepv', '-minw', 16, '-maxw', 20, '-hofract', 0.1],
  'INO_up800' => ['-dna', '-hofract', 0.2],
  'farntrans5' => ['-protein', '-maxw', 25, '-hofract', 0.2], 
  'lipocalin' => ['-protein', '-maxw', 20, '-hofract', 0.2], 
  'adh' => ['-protein', '-niter', 10, '-hofract', 0.2],
  'Klf1-200-100' => ['-dna', '-neval', 20, '-useer', '-hofract', 0.2],
  'Klf1-200' => ['-dna', '-nref', 5, '-hofract', 0.2],
  'Puf3p' => ['-rna', '-pvt', '1e-10', '-hofract', 0.2],
  'dna_with_covalent' => ['-alph', 'common/dna_with_covalent.alph', '-nmotifs', 2, '-hofract', 0.2]
);
my %obj_params = (
  'de' => ['-objfun', 'de', '-order', 1, '-nmotifs', 2],
  'cd' => ['-objfun', 'cd', '-minw', 6, '-maxw', 10, '-pvt', 0.001, '-patience', 2]
);

# extensions for each type of data
my %extensions= (
  'text' => 'txt', 
  'xml' => 'xml',
  'json' => 'html'  
);
# type of data for each extension
my %types= (
  'txt' => 'text', 
  'xml' => 'xml',
  'html' => 'json' 
);
# parts to ignore in STREME text, HTML or XML output
my %streme_ignore = (
  'text' => [
    'bioRxiv', 'Release date', 'http:', 'https:', 'MOTIFS', 'FINALTIME:', 'CPU:',
    'COMMAND:', 'Background letter frequencies'
  ], 
  'json' => [
    '^data:(version|release|cmd)', 
    '^data:runtime',
    'elapsed_time'
  ], 
  'xml' => [
    'STREME@(version|release)', 'STREME:model:command_line#value$',
    'STREME:motifs:motif@elapsed_time',
    'STREME:model:host',
    'STREME:run_time'
  ]
);

my $streme = 'streme';
my $bin = 'src';

my $counter = 1;

foreach my $obj (@objfuns) {
  my $i_dset;
  for ($i_dset=0; $i_dset<@datasets; $i_dset++) {
    my $dset = $datasets[$i_dset];
    next if ($dset eq 'Klf1-200' && !($obj eq 'cd')); 		# only CD on Klf1-200
    next if (($obj eq 'cd') && !($dset eq 'Klf1-200'));		# CD only on Klf1-200
    next if (($obj ne 'de') && ($dset eq 'dna_with_covalent'));	# DE only on dna_with_covalent
    my $reference_root = catfile('streme', 'streme.'.$dset.'.'.$obj);
    my $ext = ($dset =~ /Klf1|dna_with_covalent/) ? '.fa' : '.s';
    &test('streme'.$counter, '',
      $streme, '',
      [
	'-p', catfile('common', $dset . $ext), 
	'-oc', 'results/streme'.$counter,
	@{$obj_params{$obj}},
	@{$params{$dset}}, 
	'-verbosity', 1
      ],
      [
	{
	  output => 'results/streme'.$counter.'/streme.txt', 
	  reference => make_test_src_path($reference_root.'.txt'),
	  type => 'text',
	  ignore => $streme_ignore{'text'}
	},
	{
	  output => 'results/streme'.$counter.'/streme.xml', 
	  reference => make_test_src_path($reference_root.'.xml'),
	  type => 'xml',
	  ignore => $streme_ignore{'xml'}
	},
	{
	  output => 'results/streme'.$counter.'/streme.html', 
	  reference => make_test_src_path($reference_root.'.html'),
	  type => 'json',
	  ignore => $streme_ignore{'json'}
	},
	{
	  output => 'results/streme'.$counter.'/sequences.tsv',
	  reference => make_test_src_path($reference_root.'.sequences.tsv'),
	  type => 'text', ignore => ['^#']
	}
      ],
      catdir('results', 'streme' . $counter), $bin
    );
    $counter++;
  } # dset
} # obj