tests/testthat/test-load_taxonomyfile.R

In microresearcher/MicroVis: Microbiome Analysis and Visualization

testthat::test_that('ASV table can be cleaned with cleanASVs', {
  testthat::expect_identical(
    cleanASVs(data.frame('domain'=c('NA',
                                    'A_1','A_1','A_1','A_1','A_1','A_1','A_1'),
                         'phylum'=c('',
                                    'NA',
                                    'B_2','B_2','B_2','B_2','B_2','B_2'),
                         'class'=c('Incertae_Sedis',
                                   'NA','NA',
                                   'C_3','C_3','C_3','C_3','C_3'),
                         'order'=c('D_4',
                                   'NA','NA','NA',
                                   'D_4','D_4','D_4','D_4'),
                         'family'=c('NA',
                                    'NA','NA','NA','NA',
                                    'E_5','E_5','E_5'),
                         'genus'=c('F_6',
                                   'NA','NA','NA','NA','NA',
                                   'F_6','F_6'),
                         'species'=c('G_7',
                                     'NA','NA','NA',
                                     'NA','NA','NA',
                                     'G_7'),
                         row.names = c('NA;;Incertae_Sedis;D_4;NA;F_6;G_7',
                                       'A_1;NA;NA;NA;NA;NA;NA',
                                       'A_1;B_2;NA;NA;NA;NA;NA',
                                       'A_1;B_2;C_3;NA;NA;NA;NA',
                                       'A_1;B_2;C_3;D_4;NA;NA;NA',
                                       'A_1;B_2;C_3;D_4;E_5;NA;NA',
                                       'A_1;B_2;C_3;D_4;E_5;F_6;NA',
                                       'A_1;B_2;C_3;D_4;E_5;F_6;G_7'))),
    data.frame('domain'=c('unidentified_domain',
                          'A_1','A_1','A_1','A_1','A_1','A_1','A_1'),
               'phylum'=c('unidentified_domain_phylum',
                          'd_A_1_phylum',
                          'B_2','B_2','B_2','B_2','B_2','B_2'),
               'class'=c('unidentified_domain_class',
                         'd_A_1_class','p_B_2_class',
                         'C_3','C_3','C_3','C_3','C_3'),
               'order'=c('D_4',
                         'd_A_1_order','p_B_2_order','c_C_3_order',
                         'D_4','D_4','D_4','D_4'),
               'family'=c('o_D_4_family',
                          'd_A_1_family','p_B_2_family','c_C_3_family','o_D_4_family',
                          'E_5','E_5','E_5'),
               'genus'=c('F_6',
                         'd_A_1_genus','p_B_2_genus','c_C_3_genus','o_D_4_genus','f_E_5_genus',
                         'F_6','F_6'),
               'species'=c('F_6_G_7',
                           'd_A_1_species','p_B_2_species','c_C_3_species',
                           'o_D_4_species','f_E_5_species','g_F_6_species',
                           'F_6_G_7'),
               row.names = c('NA;;Incertae_Sedis;D_4;NA;F_6;G_7',
                             'A_1;NA;NA;NA;NA;NA;NA',
                             'A_1;B_2;NA;NA;NA;NA;NA',
                             'A_1;B_2;C_3;NA;NA;NA;NA',
                             'A_1;B_2;C_3;D_4;NA;NA;NA',
                             'A_1;B_2;C_3;D_4;E_5;NA;NA',
                             'A_1;B_2;C_3;D_4;E_5;F_6;NA',
                             'A_1;B_2;C_3;D_4;E_5;F_6;G_7')))
})

testthat::test_that('ASV table can be made from prepended taxa names', {
  invisible(capture.output(
    testthat::expect_identical(makeASVtab(c('NA;;Incertae_Sedis;D_4;NA;F_6;G_7',
                                            'k__A-1_k',
                                            'A 1;p__B:2_p',
                                            'A/1;B[_2;c__C]_3_c',
                                            'A(1;B)_2;C_3;o__D_4_o',
                                            'A_1;B_2;C_3;D_4;f__E_5_f',
                                            'A_1;B_2;C_3;D_4;E_5;g__F_6_g',
                                            'A_1;B_2;C_3;D_4;E_5;F_6;s__G_7_s')),
                               data.frame('domain'=c('unidentified_domain',
                                                     'A_1','A_1','A_1','A_1','A_1','A_1','A_1'),
                                          'phylum'=c('unidentified_domain_phylum',
                                                     'd_A_1_phylum',
                                                     'B_2','B_2','B_2','B_2','B_2','B_2'),
                                          'class'=c('unidentified_domain_class',
                                                    'd_A_1_class','p_B_2_class',
                                                    'C_3','C_3','C_3','C_3','C_3'),
                                          'order'=c('D_4',
                                                    'd_A_1_order','p_B_2_order','c_C_3_order',
                                                    'D_4','D_4','D_4','D_4'),
                                          'family'=c('o_D_4_family',
                                                     'd_A_1_family','p_B_2_family','c_C_3_family','o_D_4_family',
                                                     'E_5','E_5','E_5'),
                                          'genus'=c('F_6',
                                                    'd_A_1_genus','p_B_2_genus','c_C_3_genus','o_D_4_genus','f_E_5_genus',
                                                    'F_6','F_6'),
                                          'species'=c('F_6_G_7',
                                                      'd_A_1_species','p_B_2_species','c_C_3_species',
                                                      'o_D_4_species','f_E_5_species','g_F_6_species',
                                                      'F_6_G_7'),
                                          row.names = c('NA;;Incertae_Sedis;D_4;NA;F_6;G_7',
                                                        'A_1;NA;NA;NA;NA;NA;NA',
                                                        'A_1;B_2;NA;NA;NA;NA;NA',
                                                        'A_1;B_2;C_3;NA;NA;NA;NA',
                                                        'A_1;B_2;C_3;D_4;NA;NA;NA',
                                                        'A_1;B_2;C_3;D_4;E_5;NA;NA',
                                                        'A_1;B_2;C_3;D_4;E_5;F_6;NA',
                                                        'A_1;B_2;C_3;D_4;E_5;F_6;G_7')))))
})