annotate t/hmdb_managerTest.pl @ 29:29195cb5d378 draft default tip

master branch Updating with tag :CI_COMMIT_TAG - - Fxx

#! perl
use diagnostics;
use warnings;
no warnings qw/void/;
use strict;
no strict "refs" ;
use Test::More qw( no_plan );
use Test::Exception;
use FindBin ;
use Carp ;

## Specific Modules
use lib $FindBin::Bin ;
my $binPath = $FindBin::Bin ;
use lib::hmdbTest qw( :ALL ) ;


## To launch the right sequence : API, MAPPER, THREADER, ...
#my $sequence = 'MAPPER' ; 
my $sequence = 'MAIN' ; 
my $current_test = 1 ;

if ($sequence eq "MAIN") {
	print "\n\t\t\t\t  * * * * * * \n" ;
	print "\t  * * * - - - Test HMDB Main script - - - * * * \n\n" ;
	
	
	print "\n** Test $current_test extract_sub_mz_lists with an empty list of mzs **\n" ; $current_test++;
	
	throws_ok{ extract_sub_mz_listsTest([], 3)} '/The provided mzs list is empty/', 'Method \'extract_sub_mz_lists\' detects empty argument and died correctly' ;
	
	print "\n** Test $current_test extract_sub_mz_lists with a list of mzs and a limit of 3 **\n" ; $current_test++;
	is_deeply( extract_sub_mz_listsTest(
		['175.01', '238.19', '420.16', '780.32', '956.25', '1100.45' ], 3), 
		[ [ '175.01', '238.19', '420.16' ], [ '780.32', '956.25', '1100.45' ] ], 
		'Method \'extract_sub_mz_lists\' works with a list and return a well formated list of sublist of mzs');
	
	print "\n** Test $current_test prepare_multi_masses_query with an empty list of mzs **\n" ; $current_test++;	
	throws_ok{ prepare_multi_masses_queryTest([])} '/The input method parameter mass list is empty/', 'Method \'prepare_multi_masses_query\' detects empty argument and died correctly' ;
	
	print "\n** Test $current_test prepare_multi_masses_query with a list of mzs **\n" ; $current_test++;
	is_deeply( prepare_multi_masses_queryTest(
		['175.01', '238.19', '420.16', '780.32', '956.25', '1100.45' ] ), 
		'175.01%0D%0A238.19%0D%0A420.16%0D%0A780.32%0D%0A956.25%0D%0A1100.45%0D%0A', 
		'Method \'prepare_multi_masses_query\' works with a list of and return a well formated string for hmdb querying');
	
	print "\n** Test $current_test get_matches_from_hmdb_ua with a well-formated string of mzs **\n" ; $current_test++;
	is_deeply( getMatchesFromHmdb5WithUATest(
		'175.01%0D%0A420.16%0D%0A780.32%0D%0A956.25%0D%0A1100.45%0D%0A', 0.001, 'positive'),
		[
          'query_mass,compound_id,compound_name,formula,monoisotopic_mass,adduct,adduct_type,adduct_m/z,"delta(ppm),",ccs_value',
          '175.01,HMDB0259477,"S-(N,N-Diethyldithiocarbamoyl)-N-acetyl-L-cysteine",C10H18N2O3S3,310.04795597,M+H+K,+,175.009195,5,',
          '175.01,HMDB0258038,9-(Formyloxymethyl)-9H-fluorene-2-sulfonic acid,C15H12O5S,304.04054466,M+2Na,+,175.00949,3,',
          '175.01,HMDB0257737,"4,5-Dihydroxy-3-methyl-1H-imidazol-2-one",C4H6N2O3,130.037842061,M+2Na-H,+,175.009002,6,',
          '175.01,HMDB0240502,Dihydroresveratrol 4\'-sulfate,C14H14O6S,310.051109345,M+H+K,+,175.010772,4,',
          '175.01,HMDB0240500,Dihydroresveratrol 3-sulfate,C14H14O6S,310.051109345,M+H+K,+,175.010772,4,',
          '175.01,HMDB0246744,"5-Amino-1,3,4-thiadiazole-2-thiol",C2H3N3S2,132.976839457,M+ACN+H,+,175.010662,4,',
          '175.01,HMDB0252977,Phosphoaminophosphonic acid-guanylate ester,C10H17N6O13P3,522.006645625,M+3H,+,175.009491,3,',
          '175.01,HMDB0246800,5-Hydroxy-1-methylhydantoin,C4H6N2O3,130.037842061,M+2Na-H,+,175.009002,6,',
          '175.01,HMDB0258677,Tafamidis,C14H7Cl2NO3,306.980298509,M+ACN+2H,+,175.010699,4,',
          '175.01,HMDB0240280,Mesylate,CH4O3S,95.988114684,M+DMSO+H,+,175.009335,4,',
          '175.01,HMDB0000257,Thiosulfate,H2O3S2,113.94453531,M+IsoProp+H,+,175.009875,1,',
          '175.01,HMDB0035230,Aurantricholide B,C17H10O6,310.047738052,M+H+K,+,175.009086,5,',
          '175.01,HMDB0033657,De-O-methylsterigmatocystin,C17H10O6,310.047738052,M+H+K,+,175.009086,5,',
          '175.01,HMDB0003745,Mesna,C2H6O3S2,141.975835438,M+CH3OH+H,+,175.009324,4,',
          '420.16,HMDB0250452,Core oligosaccharide,C27H48O24,756.253552426,M+2ACN+2H,+,420.160599,1,',
          '420.16,HMDB0245742,"N-(6,8-Difluoro-2-methyl-4-quinolinyl)-N\'-[4-(dimethylamino)phenyl]urea",C19H18F2N4O,356.144867542,M+ACN+Na,+,420.160633,2,',
          '420.16,HMDB0256144,Pazopanib,C21H23N7O2S,437.163393705,M+H-H2O,+,420.160694,2,',
          '420.16,HMDB0060838,N-Desmethyl-p-hydroxyrosiglitazone,C17H17N3O4S,359.093976737,M+IsoProp+H,+,420.159317,2,',
          '420.16,HMDB0060836,N-Desmethyl-O-hydroxy rosiglitazone,C17H17N3O4S,359.093976737,M+IsoProp+H,+,420.159317,2,'
        ],
		'Method \'getMatchesFromHmdb5WithUA\' works again (API v5.0 compatible) with a well-formated string of mzs and return a complete csv from hmdb');
	
	print "\n** Test $current_test test_matches_from_hmdb_ua to get hmdb status **\n" ; $current_test++;
	is_deeply (testMatchesFromHmdb5WithUATest (), 
		\'200', 
		'The HMDB server is available: returns successful HTTP requests' ) ;
		
	print "\n** Test $current_test check_state_from_hmdb_ua to manage script execution with the hmdb server status **\n" ; $current_test++;
	is_deeply (check_state_from_hmdb_uaTest (\'200'),
		1,
		'The status 200 returns no error/warn' ) ;
		
	print "\n** Test $current_test check_state_from_hmdb_ua to manage script execution with the hmdb server status **\n" ; $current_test++;	
	throws_ok{ check_state_from_hmdb_uaTest(\'504')} 
		qr/Gateway Timeout \(504\): The HMDB server was acting as a gateway or proxy and did not receive a timely response from the upstream server\. The Hmdb tool is stopped with error\./, 
		'Method \'check_state_from_hmdb_ua\' detects HTTP error code returned by HMDB and died correctly' ;
		
		
	print "\n** Test $current_test parseHmdb5CSVResults (version 5.0) with the correct inputs for hmdb outputs parsing (csv format) **\n" ; $current_test++;
	is_deeply ( parseHmdb5CSVResultsTest (
		[
		  'query_mass,compound_id,compound_name,formula,monoisotopic_mass,adduct,adduct_type,adduct_m/z,"delta(ppm),",ccs_value',
          '175.01,HMDB0259477,"S-(N,N-Diethyldithiocarbamoyl)-N-acetyl-L-cysteine",C10H18N2O3S3,310.04795597,M+H+K,+,175.009195,5,',
          '175.01,HMDB0258038,9-(Formyloxymethyl)-9H-fluorene-2-sulfonic acid,C15H12O5S,304.04054466,M+2Na,+,175.00949,3,',
          '175.01,HMDB0257737,"4,5-Dihydroxy-3-methyl-1H-imidazol-2-one",C4H6N2O3,130.037842061,M+2Na-H,+,175.009002,6,',
          '175.01,HMDB0240502,Dihydroresveratrol 4\'-sulfate,C14H14O6S,310.051109345,M+H+K,+,175.010772,4,',
          '175.01,HMDB0240500,Dihydroresveratrol 3-sulfate,C14H14O6S,310.051109345,M+H+K,+,175.010772,4,',
          '175.01,HMDB0246744,"5-Amino-1,3,4-thiadiazole-2-thiol",C2H3N3S2,132.976839457,M+ACN+H,+,175.010662,4,',
          '175.01,HMDB0252977,Phosphoaminophosphonic acid-guanylate ester,C10H17N6O13P3,522.006645625,M+3H,+,175.009491,3,',
          '175.01,HMDB0246800,5-Hydroxy-1-methylhydantoin,C4H6N2O3,130.037842061,M+2Na-H,+,175.009002,6,',
          '175.01,HMDB0258677,Tafamidis,C14H7Cl2NO3,306.980298509,M+ACN+2H,+,175.010699,4,',
          '175.01,HMDB0240280,Mesylate,CH4O3S,95.988114684,M+DMSO+H,+,175.009335,4,',
          '175.01,HMDB0000257,Thiosulfate,H2O3S2,113.94453531,M+IsoProp+H,+,175.009875,1,',
          '175.01,HMDB0035230,Aurantricholide B,C17H10O6,310.047738052,M+H+K,+,175.009086,5,',
          '175.01,HMDB0033657,De-O-methylsterigmatocystin,C17H10O6,310.047738052,M+H+K,+,175.009086,5,',
          '175.01,HMDB0003745,Mesna,C2H6O3S2,141.975835438,M+CH3OH+H,+,175.009324,4,',
          '420.16,HMDB0250452,Core oligosaccharide,C27H48O24,756.253552426,M+2ACN+2H,+,420.160599,1,',
          '420.16,HMDB0245742,"N-(6,8-Difluoro-2-methyl-4-quinolinyl)-N\'-[4-(dimethylamino)phenyl]urea",C19H18F2N4O,356.144867542,M+ACN+Na,+,420.160633,2,',
          '420.16,HMDB0256144,Pazopanib,C21H23N7O2S,437.163393705,M+H-H2O,+,420.160694,2,',
          '420.16,HMDB0060838,N-Desmethyl-p-hydroxyrosiglitazone,C17H17N3O4S,359.093976737,M+IsoProp+H,+,420.159317,2,',
          '420.16,HMDB0060836,N-Desmethyl-O-hydroxy rosiglitazone,C17H17N3O4S,359.093976737,M+IsoProp+H,+,420.159317,2,'
        ],
        ['175.01', '238.19', '420.16'],
        10
	),
		[
          [
            { 'ENTRY_CPD_MZ' => '310.04795597', 'ENTRY_DELTA' => '5', 'ENTRY_FORMULA' => 'C10H18N2O3S3', 'ENTRY_ADDUCT' => 'M+H+K', 'ENTRY_ENTRY_NAME' => '"S-(N,N-Diethyldithiocarbamoyl)-N-acetyl-L-cysteine"', 'ENTRY_ENTRY_ID' => 'HMDB0259477', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ADDUCT_MZ' => '175.009195' },
            { 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ADDUCT_MZ' => '175.00949', 'ENTRY_ENTRY_ID' => 'HMDB0258038', 'ENTRY_ENTRY_NAME' => '9-(Formyloxymethyl)-9H-fluorene-2-sulfonic acid', 'ENTRY_FORMULA' => 'C15H12O5S', 'ENTRY_ADDUCT' => 'M+2Na', 'ENTRY_DELTA' => '3', 'ENTRY_CPD_MZ' => '304.04054466' },
            { 'ENTRY_FORMULA' => 'C14H14O6S', 'ENTRY_ADDUCT' => 'M+H+K', 'ENTRY_ENTRY_NAME' => 'Dihydroresveratrol 4\'-sulfate', 'ENTRY_CPD_MZ' => '310.051109345', 'ENTRY_DELTA' => '4', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ADDUCT_MZ' => '175.010772', 'ENTRY_ENTRY_ID' => 'HMDB0240502' },
            { 'ENTRY_ADDUCT_MZ' => '175.010772', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ENTRY_ID' => 'HMDB0240500', 'ENTRY_ADDUCT' => 'M+H+K', 'ENTRY_FORMULA' => 'C14H14O6S', 'ENTRY_ENTRY_NAME' => 'Dihydroresveratrol 3-sulfate', 'ENTRY_CPD_MZ' => '310.051109345', 'ENTRY_DELTA' => '4' },
            { 'ENTRY_ENTRY_ID' => 'HMDB0246744', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ADDUCT_MZ' => '175.010662', 'ENTRY_CPD_MZ' => '132.976839457', 'ENTRY_DELTA' => '4', 'ENTRY_ENTRY_NAME' => '"5-Amino-1,3,4-thiadiazole-2-thiol"', 'ENTRY_FORMULA' => 'C2H3N3S2', 'ENTRY_ADDUCT' => 'M+ACN+H' },
            { 'ENTRY_ENTRY_ID' => 'HMDB0252977', 'ENTRY_ADDUCT_MZ' => '175.009491', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_CPD_MZ' => '522.006645625', 'ENTRY_DELTA' => '3', 'ENTRY_ADDUCT' => 'M+3H', 'ENTRY_FORMULA' => 'C10H17N6O13P3', 'ENTRY_ENTRY_NAME' => 'Phosphoaminophosphonic acid-guanylate ester' },
            { 'ENTRY_ENTRY_ID' => 'HMDB0258677', 'ENTRY_ADDUCT_MZ' => '175.010699', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_CPD_MZ' => '306.980298509', 'ENTRY_DELTA' => '4', 'ENTRY_ADDUCT' => 'M+ACN+2H', 'ENTRY_FORMULA' => 'C14H7Cl2NO3', 'ENTRY_ENTRY_NAME' => 'Tafamidis' },
            { 'ENTRY_ENTRY_ID' => 'HMDB0240280', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ADDUCT_MZ' => '175.009335', 'ENTRY_DELTA' => '4', 'ENTRY_CPD_MZ' => '95.988114684', 'ENTRY_FORMULA' => 'CH4O3S', 'ENTRY_ADDUCT' => 'M+DMSO+H', 'ENTRY_ENTRY_NAME' => 'Mesylate' },
            { 'ENTRY_DELTA' => '1', 'ENTRY_CPD_MZ' => '113.94453531', 'ENTRY_FORMULA' => 'H2O3S2', 'ENTRY_ADDUCT' => 'M+IsoProp+H', 'ENTRY_ENTRY_NAME' => 'Thiosulfate', 'ENTRY_ENTRY_ID' => 'HMDB0000257', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ADDUCT_MZ' => '175.009875' },
            { 'ENTRY_DELTA' => '5', 'ENTRY_CPD_MZ' => '310.047738052', 'ENTRY_ENTRY_NAME' => 'Aurantricholide B', 'ENTRY_FORMULA' => 'C17H10O6', 'ENTRY_ADDUCT' => 'M+H+K', 'ENTRY_ENTRY_ID' => 'HMDB0035230', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ADDUCT_MZ' => '175.009086' }
          ],
          [],
          [
            { 'ENTRY_ADDUCT_MZ' => '420.160599', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ENTRY_ID' => 'HMDB0250452', 'ENTRY_ADDUCT' => 'M+2ACN+2H', 'ENTRY_FORMULA' => 'C27H48O24', 'ENTRY_ENTRY_NAME' => 'Core oligosaccharide', 'ENTRY_CPD_MZ' => '756.253552426', 'ENTRY_DELTA' => '1' },
            { 'ENTRY_CPD_MZ' => '356.144867542', 'ENTRY_DELTA' => '2', 'ENTRY_ENTRY_NAME' => '"N-(6,8-Difluoro-2-methyl-4-quinolinyl)-N\'-[4-(dimethylamino)phenyl]urea"', 'ENTRY_FORMULA' => 'C19H18F2N4O', 'ENTRY_ADDUCT' => 'M+ACN+Na', 'ENTRY_ENTRY_ID' => 'HMDB0245742', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ADDUCT_MZ' => '420.160633' },
            { 'ENTRY_ADDUCT_MZ' => '420.159317', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ENTRY_ID' => 'HMDB0060838', 'ENTRY_ADDUCT' => 'M+IsoProp+H', 'ENTRY_FORMULA' => 'C17H17N3O4S', 'ENTRY_ENTRY_NAME' => 'N-Desmethyl-p-hydroxyrosiglitazone', 'ENTRY_CPD_MZ' => '359.093976737', 'ENTRY_DELTA' => '2' },
            { 'ENTRY_ADDUCT_MZ' => '420.159317', 'ENTRY_ADDUCT_TYPE' => '+', 'ENTRY_ENTRY_ID' => 'HMDB0060836', 'ENTRY_ADDUCT' => 'M+IsoProp+H', 'ENTRY_FORMULA' => 'C17H17N3O4S', 'ENTRY_ENTRY_NAME' => 'N-Desmethyl-O-hydroxy rosiglitazone', 'ENTRY_DELTA' => '2', 'ENTRY_CPD_MZ' => '359.093976737' }
          ]
        ],
		'Method \'parseHmdb5CSVResults\' works with a well-formated csv output and returns a a well formated array' ) ;


#	print "\n** Test $current_test parse_hmdb_csv_results with a void hmdb output and a list of mzs **\n" ; $current_test++;
#	is_deeply ( parse_hmdb_csv_resultsTest ( [], ['175.01', '238.19', '420.16'] ),
#		[ [], [], [] ],
#		'Method \'parse_hmdb_csv_results\' works with a empty csv output and returns an empty but well formatted array' ) ;
#	
#	print "\n** Test $current_test parse_hmdb_csv_results with a void hmdb output and a void mz list  **\n" ; $current_test++;
#	is_deeply ( parse_hmdb_csv_resultsTest ( [], [] ),
#		[],
#		'Method \'parse_hmdb_csv_results\' works with a empty csv output/mz list and returns an empty but well formatted array' ) ;
#
#	## Other example... from mz 265.0567515
#	print "\n** Test $current_test parse_hmdb_csv_results with mz 265.0567515 and the correct inputs for hmdb outputs parsing (csv format) **\n" ; $current_test++;
#	is_deeply ( parse_hmdb_csv_resultsTest (
#		[
#        'query_mass,compound_id,compound_name,kegg_id,formula,monoisotopic_mass,adduct,adduct_type,adduct_m/z,delta(ppm)',
#		'265.0567515,HMDB0059667,3-Methylsulfolene,n/a,C5H8O2S,132.02450019,2M+H,+,265.056276,2',
#		'265.0567515,HMDB0038767,Camelliaside b,n/a,C32H38O19,726.200729034,M+3Na,+,265.056128,2',
#		'265.0567515,HMDB0039759,Kaempferol 3-(2g-apiosylrobinobioside),n/a,C32H38O19,726.200729034,M+3Na,+,265.056128,2',
#		'265.0567515,HMDB0040875,Isoschaftoside 4\'-glucoside,n/a,C32H38O19,726.200729034,M+3Na,+,265.056128,2',
#		'265.0567515,HMDB0040878,Schaftoside 4\'-glucoside,n/a,C32H38O19,726.200729034,M+3Na,+,265.056128,2',
#		'265.0567515,HMDB0059600,erythro-5-Phosphonooxy-L-lysine,C03366,C6H15N2O6P,242.066772734,M+Na,+,265.055991,3',
#		'265.0567515,HMDB0061174,"2,4-Imidazolidinedione, 3-ethyl-5-(4-hydroxyphenyl)-",n/a,C11H12N2O3,220.08479226,M+2Na-H,+,265.055952,3',
#		'265.0567515,HMDB0015571,Oxitriptan,C00643,C11H12N2O3,220.08479226,M+2Na-H,+,265.055952,3',
#		'265.0567515,HMDB0000472,5-hydroxy-l-tryptophan,C01017,C11H12N2O3,220.08479226,M+2Na-H,+,265.055952,3',
#		'265.0567515,HMDB0030584,Silidianin,n/a,C25H24O10,484.136946988,M+2Na,+,265.057691,4',
#        ],
#        ['265.0567515'],
#        5
#	),
#		[
#          [
#            {
#              'ENTRY_DELTA' => '2',
#              'ENTRY_FORMULA' => 'C5H8O2S',
#              'ENTRY_ENTRY_ID' => 'HMDB0059667',
#              'ENTRY_ADDUCT_MZ' => '265.056276',
#              'ENTRY_ADDUCT_TYPE' => '+',
#              'ENTRY_CPD_MZ' => '132.02450019',
#              'ENTRY_ADDUCT' => '2M+H'
#            },
#            {
#              'ENTRY_ADDUCT_MZ' => '265.056128',
#              'ENTRY_ENTRY_ID' => 'HMDB0038767',
#              'ENTRY_ADDUCT_TYPE' => '+',
#              'ENTRY_CPD_MZ' => '726.200729034',
#              'ENTRY_ADDUCT' => 'M+3Na',
#              'ENTRY_DELTA' => '2',
#              'ENTRY_FORMULA' => 'C32H38O19'
#            },
#            {
#              'ENTRY_DELTA' => '2',
#              'ENTRY_FORMULA' => 'C32H38O19',
#              'ENTRY_ENTRY_ID' => 'HMDB0039759',
#              'ENTRY_ADDUCT_MZ' => '265.056128',
#              'ENTRY_ADDUCT_TYPE' => '+',
#              'ENTRY_CPD_MZ' => '726.200729034',
#              'ENTRY_ADDUCT' => 'M+3Na'
#            },
#            {
#              'ENTRY_FORMULA' => 'C32H38O19',
#              'ENTRY_DELTA' => '2',
#              'ENTRY_ADDUCT_TYPE' => '+',
#              'ENTRY_ADDUCT_MZ' => '265.056128',
#              'ENTRY_ENTRY_ID' => 'HMDB0040875',
#              'ENTRY_CPD_MZ' => '726.200729034',
#              'ENTRY_ADDUCT' => 'M+3Na'
#            },
#            {
#              'ENTRY_ADDUCT_TYPE' => '+',
#              'ENTRY_ENTRY_ID' => 'HMDB0040878',
#              'ENTRY_ADDUCT_MZ' => '265.056128',
#              'ENTRY_ADDUCT' => 'M+3Na',
#              'ENTRY_CPD_MZ' => '726.200729034',
#              'ENTRY_FORMULA' => 'C32H38O19',
#              'ENTRY_DELTA' => '2'
#            }
#          ]
#        ],
#		'Method \'parse_hmdb_csv_results\' works with a well-formated csv output and returns a a well formated array' ) ;	
#		
#	

}
## END of the script