lib/CXGN/Phylo/Species_name_map.pm

   1 use strict;
   2
   3 =head1 NAME
   4
   5 CXGN::Phylo::Species_name_map - a package that handles variant species names (e.g. tomato, Lycopersicon_esculentum, Solanum_lycopersicum)
   6 =head1 DESCRIPTION
   7
   8         This is for handling different species names that may be used. There is a  hash whose keys are variants, and
   9         values are the standard versions of the species names.
  10         A default hash is set up in new. You can add other variant/standard pairs. The keys are all
  11         in a standard format with words separated by _, initial letter uc, others lc. e.g. if $sname = POTATO
  12         get_standard_name($sname) will transform this to Potato, and return the value corresponding to this key,
  13         which would be Solanum_tuberosum.
  14
  15 =head1 AUTHOR
  16
  17 Tom York (tly2@cornell.edu)
  18
  19 =head1 FUNCTIONS
  20
  21         This class implements the following functions:
  22
  23 =cut
  24
  25                 package  CXGN::Phylo::Species_name_map;
  26 # this is a class to map variant species names to standard species names
  27
  28 =head2 function new()
  29
  30         Synopsis:       $snm->CXGN::Phylo::Species_name_map->new();
  31 Arguments:      none.
  32 Returns:        ref to newly constructed Species_name_map object with a default set of key/value (species/std species) pairs.
  33 Description: The objects hash has a key name_hash, whose corresponding value is a reference to a hash whose
  34 keys are variant species names and values are standard species names. Both the keys and values of name_hash are
  35 in a standard format using _ as separator, and all lowercase except first char. which is uppercase (see to_standard_format).
  36
  37 =cut
  38
  39 sub new{
  40         my $class = shift;
  41         my $args = {};
  42         my $self = bless $args, $class;
  43
  44         %{$self->{name_hash}} = ();
  45 #list of default standard names
  46         my @std_species = (
  47                         'Solanum_lycopersicum',
  48                         'Solanum_tuberosum',
  49                         'Solanum_melongena',
  50                         'Capsicum_annuum',
  51                         'Nicotiana_tabacum',
  52                         'Petunia',
  53                         'Coffea_arabica',
  54                         'Coffea_canephora',
  55                         'Antirrhinum',
  56                         'Arabidopsis_thaliana',
  57                         'Ipomoea_batatas',
  58                         'Oryza_sativa',
  59                         'Brachypodium_distachyon',
  60                         'Medicago_truncatula',
  61                         'Glycine_max',
  62                         'Zea_mays',
  63                         'Sorghum_bicolor',
  64                         'Carica_papaya',
  65                         'Vitis_vinifera',
  66                         'Ricinus_communis',
  67                         'Populus_trichocarpa',
  68                         'Selaginella_moellendorffii',
  69
  70                         'Chlamydomonas_reinhardtii',
  71                         'Physcomitrella_patens',
  72                         'Pinus_taeda',
  73                         'Amborella_trichopoda',
  74                         'Phoenix_dactylifera',
  75                         'Setaria_italica',
  76                         'Triticum_aestivum',
  77                         'Hordeum_vulgare',
  78                         'Aquilegia_coerulea',
  79                         'Helianthus_annuus',
  80                         'Lactuca_sativa',
  81                         'Daucus_carota',
  82                         'Eucalyptus_grandis',
  83                         'Citrus_sinensis',
  84                         'Citrus_clementina',
  85                         'Theobroma_cacao',
  86                         'Gossypium_raimondii',
  87                         'Brassica_rapa',
  88                         'Thellungiella_parvula',
  89                         'Capsella_rubella',
  90                         'Arabidopsis_lyrata',
  91                         'Arachis_hypogaea',
  92                         'Cajanus_cajan',
  93                         'Lotus_japonica',
  94                         'Cannabis_sativa',
  95                         'Malus_domestica',
  96                         'Prunus_persica',
  97                         'Fragaria_vesca',
  98                         'Cucumis_sativus',
  99                         'Manihot_esculenta',
 100                         'Linum_usitatissimum'
 101                                 );
 102
 103 # first set up some standard name associations
 104         foreach my $s (@std_species) {
 105                 $self->set_standard_name($s, $s);
 106         }
 107
 108         $self->set_standard_name('tomato', 'Solanum_lycopersicum');
 109         $self->set_standard_name('potato', 'Solanum_tuberosum');
 110         $self->set_standard_name('eggplant', 'Solanum_melongena');
 111         $self->set_standard_name('pepper', 'Capsicum_annuum');
 112         $self->set_standard_name('tobacco', 'Nicotiana_tabacum');
 113         $self->set_standard_name('petunia', 'Petunia'); # species?
 114                 $self->set_standard_name('sweet_potato', 'Ipomoea_batatas');
 115         $self->set_standard_name('coffee', 'Coffea_arabica'); #what about C. canephora?
 116                 $self->set_standard_name('arabica_coffee', 'Coffea_arabica'); #what about C. canephora?
 117                 $self->set_standard_name('robusta_coffee', 'Coffea_canephora');
 118         $self->set_standard_name('rice', 'Oryza_sativa');
 119         $self->set_standard_name('brachypodium', 'Brachypodium_distachyon');
 120
 121         $self->set_standard_name('snapdragon', 'Antirrhinum'); #species Majus?
 122                 $self->set_standard_name('arabidopsis', 'Arabidopsis_thaliana');
 123
 124         $self->set_standard_name('Solanum betaceum', 'Solanum_betaceum');
 125         $self->set_standard_name('tamarillo', 'Solanum_betaceum');
 126
 127         $self->set_standard_name('Physalis philadelphica', 'Physalis_philadelphica');
 128         $self->set_standard_name('tomatillo', 'Physalis_philadelphica');
 129
 130         $self->set_standard_name('Lycopersicon Esculentum', $self->get_standard_name('tomato'));
 131         $self->set_standard_name('selaginella', 'Selaginella_moellendorffii');
 132         $self->set_standard_name('maize', 'Zea_mays');
 133         $self->set_standard_name('sorghum', 'Sorghum_bicolor');
 134         $self->set_standard_name('castorbean', 'Ricinus_communis');
 135         $self->set_standard_name('grape', 'Vitis_vinifera');
 136         $self->set_standard_name('papaya', 'Carica_papaya');
 137         $self->set_standard_name('soy', 'Glycine_max');
 138         $self->set_standard_name('soybean', 'Glycine_max');
 139         $self->set_standard_name('medicago', 'Medicago_truncatula');
 140         $self->set_standard_name('poplar', 'Populus_trichocarpa');
 141
 142         $self->set_standard_name('chlamydomonas', 'Chlamydomonas_reinhardtii');
 143         $self->set_standard_name('physcomitrella', 'Physcomitrella_patens');
 144         $self->set_standard_name('loblolly_pine', 'Pinus_taeda');
 145         $self->set_standard_name('amborella', 'Amborella_trichopoda');
 146         $self->set_standard_name('foxtail_millet', 'Setaria_italica');
 147         $self->set_standard_name('date_palm', 'Phoenix_dactylifera');
 148         $self->set_standard_name('wheat', 'Triticum_aestivum');
 149         $self->set_standard_name('barley', 'Hordeum_vulgare');
 150
 151         $self->set_standard_name('peach', 'Prunus_persica');
 152         $self->set_standard_name('peanut', 'Arachis_hypogaea');
 153         $self->set_standard_name('pigeon_pea', 'Cajanus_cajan');
 154         $self->set_standard_name('lotus', 'Lotus_japonica');
 155         $self->set_standard_name('apple', 'Malus_domestica');
 156         $self->set_standard_name('cannabis', 'Cannabis_sativa');
 157         $self->set_standard_name('hemp', 'Cannabis_sativa');
 158         $self->set_standard_name('woodland_strawberry', 'Fragaria_vesca');
 159         $self->set_standard_name('cucumber', 'Cucumis_sativus');
 160         $self->set_standard_name('flax', 'Linum_usitatissimum');
 161         $self->set_standard_name('cassava', 'Manihot_esculenta');
 162 # print STDOUT "in CXGN::Phylo::Species_name_map->new(). tomato std name: ", $self->get_standard_name("tomato"), "\n";
 163         return $self;
 164 }
 165
 166 =head2 function set_standard_name
 167
 168 Synopsis:        $snm->set_standard_name("tomatillo", "Physalis philadelphica")
 169 Arguments: List of two strings; the second becomes the standard species name corresponding to the first.
 170 Returns:        nothing
 171 Side effects: Stores a key value pair in the hash
 172 Description:     Both arguments are transformed by to_standard_format; these transformed args become a key-value pair of name_hash.
 173
 174 =cut
 175
 176 sub set_standard_name{
 177         my $self=shift;
 178         my $var = shift;
 179         my $std = shift;
 180         $var = CXGN::Phylo::Species_name_map->to_standard_format($var); #so hash keys are in standard format e.g. " solanum  LYCOPERSICUM " ->  "Solanum_lycopersicum"
 181                 $std = CXGN::Phylo::Species_name_map->to_standard_format($std); # so hash vals are in standard format
 182                 $self->{name_hash}->{$var} = $std;
 183 }
 184
 185 =head2 function get_standard_name
 186
 187 Synopsis:        my $std_species = $snm->get_standard_name("tomatillo")
 188 Arguments: A species name string;
 189 Returns:        The corresponding standard name.
 190 Description:     The argument is transformed by to_standard_format, and this becomes a key for which the value is returned.
 191
 192 =cut
 193
 194 sub get_standard_name{
 195         my $self = shift;
 196         my $var = shift;
 197         $var = CXGN::Phylo::Species_name_map->to_standard_format($var); #e.g. " solanum  LYCOPERSICUM " ->  "Solanum_lycopersicum"
 198                 return $self->{name_hash}->{$var};
 199 }
 200
 201 =head2 function copy()
 202
 203 Synopsis:        my $snm_copy = $snm->copy()
 204         Arguments:      a Species_name_map object
 205         Returns:        a Species_name_map object, a copy of $snm
 206         Description:     Starts with a new default object, and copies the  hash to it
 207
 208 =cut
 209
 210 sub copy{                                                                                               # just copy the hash
 211         my $self = shift;
 212         my $new = CXGN::Phylo::Species_name_map->new();
 213         foreach my $k (keys %$self) {
 214                 $new->set_standard_name($k, $self->get_standard_name($k));
 215         }
 216         return $new;
 217 }
 218
 219 =head2 function to_standard_format
 220
 221 Synopsis:        my $str = CXGN::Phylo::Species_name_map->to_standard_format($sp_name);
 222 Arguments:      A string.
 223 Returns:        A string put into a standard format by removing leading and trailing whitespace, splitting at whitespace or _,
 224         making lc, joining with _, then make first char uc. E.g.: "   solanum _TUBEROSUM  "  becomes "Solanum_tuberosum".
 225         Note that applying this multiple times gives same result as applying once.
 226         Description:     The idea is to have all the hash keys of a Species_name_map be in this format, so that some minor variations
 227         (e.g. potato, Potato) would share the same key
 228
 229 =cut
 230
 231 sub to_standard_format{                                 # remove initial, final whitespace,  replace whitespace and _ with single space separating pieces which are ucfirst lc
 232         my $self = shift;
 233         my $species = shift;
 234         if (defined $species) {
 235                 $species =~ s/^\s+//;                       # remove initial whitespace
 236                         $species =~ s/\s+$//;                       # remove final whitespace
 237                         my @word_list = split(/[\s_]+/, $species);  # split on _, whitespace
 238                         map($_ = lc $_, @word_list);                # -> all lowercase
 239                         $species = join("_", @word_list);           # join with _'s
 240                         $species = ucfirst $species;                # Capitalize first letter (genus name)
 241         }
 242         return $species;
 243 }
 244
 245 sub to_display_format{
 246         my $self = shift;
 247         my $species = shift;
 248         $species =~ s/_/ /g;
 249         return $species;
 250 }
 251
 252 =head2 function get_map_string
 253
 254 Synopsis:       my $str = $snm->get_map_string()
 255         Arguments:      a Species_name_map object
 256         Returns:        A string with the keys and values (standard species) of the hash in form:
 257 (potato => Solanum_tuberosum, tomato => Solanum_lycopersicum)
 258         Description:
 259
 260 =cut
 261
 262 sub get_map_string{
 263         my $self = shift;
 264         my $string = "(";
 265         foreach my $s (keys %{$self->{name_hash}}) {
 266                 $string .= $s . " => " . $self->get_standard_name($s) . ",  ";
 267         }
 268         $string =~ s/,\s*$//g; # eliminate final comma and whitespace
 269                 $string .= ")";
 270         return $string;
 271 }
 272
 273
 274 1;