# One-off script prepared to provide data to David Hibbett and Romina Gazis.
# Lists numbers of species in each fungal order.
from org.opentreeoflife.taxa import Taxonomy
import csv, sys
from taxonomies import load_fung, load_ncbi, load_gbif, load_irmng
taxonomies = [('fung', load_fung(), 'Index Fungorum'),
('ncbi', load_ncbi(), 'NCBI'),
('gbif', load_gbif(), 'GBIF'),
('irmng', load_irmng(), 'IRMNG'),
('ott', Taxonomy.getTaxonomy('tax/ott/'), 'OTT 2.9'),
]
def main():
infile = open('order-counts-orders.csv', 'r')
reader = csv.reader(infile)
reader.next() #header row
taxa = ['Fungi']
for tuple in reader:
taxa.append(tuple[0])
infile.close()
write_counts(taxa)
def write_counts(taxa):
outfile = open('order-counts.csv', 'w')
writer = csv.writer(outfile)
header = ['order']
for (name, taxonomy, label) in taxonomies:
header += [label + ' bin', label + ' sp', label + ' tip']
def prepare_ncbi(ott):
ncbi = taxonomies.load_ncbi()
ott.addSource(ncbi)
# David Hibbett has requested that for Fungi, only Index Fungorum
# should be seen. Rather than delete the NCBI fungal taxa, we just
# mark them 'hidden' so they can be suppressed downstream. This
# preserves the identifier assignments, which may have been used
# somewhere.
ncbi.taxon('Fungi').hideDescendantsToRank('species')
# - Alignment to OTT -
#ott.same(ncbi.taxon('Cyanobacteria'), silva.taxon('D88288/#3'))
# #### Check - was fungi.taxon
# ** No unique taxon found with this name: Burkea
# ** No unique taxon found with this name: Coscinium
# ** No unique taxon found with this name: Perezia
# ott.notSame(ncbi.taxon('Burkea', 'Viridiplantae'), ott.taxon('Burkea'))
# ott.notSame(ncbi.taxon('Coscinium', 'Viridiplantae'), ott.taxon('Coscinium'))
# ott.notSame(ncbi.taxon('Perezia', 'Viridiplantae'), ott.taxon('Perezia'))
# JAR 2014-04-11 Discovered during regression testing
# now handled in other ways
# ott.notSame(ncbi.taxon('Epiphloea', 'Rhodophyta'), ott.taxon('Epiphloea', 'Ascomycota'))
# JAR attempt to resolve ambiguous alignment of Trichosporon in IF and
# NCBI based on common parent and member.
# Type = T. beigelii, which is current, according to Mycobank.
# But I'm going to use a different 'type', Trichosporon cutaneum.
# #### Check - was fungi.taxon
ott.same(ott.taxonThatContains('Trichosporon', 'Trichosporon cutaneum'),
#ncbi.taxonThatContains('Trichosporon', 'Trichosporon cutaneum')
ncbi.taxon('5552')
)
# 2014-04-23 In new version of IF - obvious misalignment
# #### Check - was fungi.taxon
# ott.notSame(ncbi.taxon('Crepidula', 'Gastropoda'), ott.taxon('Crepidula', 'Microsporidia'))
# ott.notSame(ncbi.taxon('Hessea', 'Viridiplantae'), ott.taxon('Hessea', 'Microsporidia'))
# 2014-04-23 Resolve ambiguity introduced into new version of IF
# http://www.speciesfungorum.org/Names/SynSpecies.asp?RecordID=331593
# #### Check - was fungi.taxon
ott.same(ncbi.taxon('Gymnopilus spectabilis var. junonius'), ott.taxon('Gymnopilus junonius'))
# JAR 2014-04-23 More sample contamination in SILVA 115
# #### Check - was fungi.taxon
# ott.same(ncbi.taxon('Lamprospora'), ott.taxon('Lamprospora', 'Pyronemataceae'))
# JAR 2014-04-25
# ### CHECK: was silva.taxon
# ott.notSame(ott.taxon('Bostrychia', 'Rhodophyceae'), ncbi.taxon('Bostrychia', 'Aves'))
# https://github.com/OpenTreeOfLife/feedback/issues/45
if False:
ott.notSame(ott.maybeTaxon('Choanoflagellida', 'Ichthyosporea'),
ncbi.maybeTaxon('Choanoflagellida', 'Opisthokonta'))
# Dail 2014-03-31 https://github.com/OpenTreeOfLife/feedback/issues/5
# updated 2015-06-28 NCBI Katablepharidophyta = SILVA Kathablepharidae.
# ### CHECK: was silva.taxon
ott.same(ncbi.taxon('Katablepharidophyta'), ott.taxon('Kathablepharidae'))
# was: ott.taxon('Katablepharidophyta').hide()
# probably not needed
ott.same(ncbi.taxon('Ciliophora', 'Alveolata'), ott.taxon('Ciliophora', 'Alveolata'))
ott.notSame(ncbi.taxon('Diphylleia', 'Chloroplastida'),
ott.taxonThatContains('Diphylleia', 'Diphylleia rotans'))
# 2015-10-06 JAR noticed while debugging surprisingly large number of newly-hidden ids
# n.b. there are three Ctenophoras: comb jellies, diatom, fly
ott.notSame(ncbi.taxon('Ctenophora', 'Metazoa'),
ott.taxon('Ctenophora', 'Stramenopiles')) # diatom from SILVA, WoRMS, GBIF, IRMNG
return ncbi
# One-off script prepared to provide data to David Hibbett and Romina Gazis.
# Lists numbers of species in each fungal order.
from org.opentreeoflife.smasher import Taxonomy
import csv, sys
from taxonomies import load_fung, load_ncbi, load_gbif, load_irmng
taxonomies = [
("fung", load_fung(), "Index Fungorum"),
("ncbi", load_ncbi(), "NCBI"),
("gbif", load_gbif(), "GBIF"),
("irmng", load_irmng(), "IRMNG"),
("ott", Taxonomy.getTaxonomy("tax/ott/"), "OTT 2.9"),
]
def main():
infile = open("order-counts-orders.csv", "r")
reader = csv.reader(infile)
reader.next() # header row
taxa = ["Fungi"]
for tuple in reader:
taxa.append(tuple[0])
infile.close()
write_counts(taxa)
def write_counts(taxa):
outfile = open("order-counts.csv", "w")
writer = csv.writer(outfile)
def prepare_ncbi(ott):
ncbi = taxonomies.load_ncbi()
# David Hibbett has requested that for Fungi, only Index Fungorum
# should be seen. Rather than delete the NCBI fungal taxa, we just
# mark them 'hidden' so they can be suppressed downstream. This
# preserves the identifier assignments, which may have been used
# somewhere.
ncbi.taxon('Fungi').hideDescendantsToRank('species')
# - Alignment to OTT -
#ott.same(ncbi.taxon('Cyanobacteria'), silva.taxon('D88288/#3'))
# #### Check - was fungi.taxon
# ** No unique taxon found with this name: Burkea
# ** No unique taxon found with this name: Coscinium
# ** No unique taxon found with this name: Perezia
# ott.notSame(ncbi.taxon('Burkea', 'Viridiplantae'), ott.taxon('Burkea'))
# ott.notSame(ncbi.taxon('Coscinium', 'Viridiplantae'), ott.taxon('Coscinium'))
# ott.notSame(ncbi.taxon('Perezia', 'Viridiplantae'), ott.taxon('Perezia'))
# JAR 2014-04-11 Discovered during regression testing
# now handled in other ways
# ott.notSame(ncbi.taxon('Epiphloea', 'Rhodophyta'), ott.taxon('Epiphloea', 'Ascomycota'))
# JAR attempt to resolve ambiguous alignment of Trichosporon in IF and
# NCBI based on common parent and member.
# Type = T. beigelii, which is current, according to Mycobank.
# But I'm going to use a different 'type', Trichosporon cutaneum.
# #### Check - was fungi.taxon
ott.same(
ott.taxonThatContains('Trichosporon', 'Trichosporon cutaneum'),
#ncbi.taxonThatContains('Trichosporon', 'Trichosporon cutaneum')
ncbi.taxon('5552'))
# 2014-04-23 In new version of IF - obvious misalignment
# #### Check - was fungi.taxon
# ott.notSame(ncbi.taxon('Crepidula', 'Gastropoda'), ott.taxon('Crepidula', 'Microsporidia'))
# ott.notSame(ncbi.taxon('Hessea', 'Viridiplantae'), ott.taxon('Hessea', 'Microsporidia'))
# 2014-04-23 Resolve ambiguity introduced into new version of IF
# http://www.speciesfungorum.org/Names/SynSpecies.asp?RecordID=331593
# #### Check - was fungi.taxon
ott.same(ncbi.taxon('Gymnopilus spectabilis var. junonius'),
ott.taxon('Gymnopilus junonius'))
# JAR 2014-04-23 More sample contamination in SILVA 115
# #### Check - was fungi.taxon
# ott.same(ncbi.taxon('Lamprospora'), ott.taxon('Lamprospora', 'Pyronemataceae'))
# JAR 2014-04-25
# ### CHECK: was silva.taxon
# ott.notSame(ott.taxon('Bostrychia', 'Rhodophyceae'), ncbi.taxon('Bostrychia', 'Aves'))
# https://github.com/OpenTreeOfLife/feedback/issues/45
if False:
ott.notSame(ott.maybeTaxon('Choanoflagellida', 'Ichthyosporea'),
ncbi.maybeTaxon('Choanoflagellida', 'Opisthokonta'))
# Dail 2014-03-31 https://github.com/OpenTreeOfLife/feedback/issues/5
# updated 2015-06-28 NCBI Katablepharidophyta = SILVA Kathablepharidae.
# ### CHECK: was silva.taxon
ott.same(ncbi.taxon('Katablepharidophyta'), ott.taxon('Kathablepharidae'))
# was: ott.taxon('Katablepharidophyta').hide()
# probably not needed
ott.same(ncbi.taxon('Ciliophora', 'Alveolata'),
ott.taxon('Ciliophora', 'Alveolata'))
ott.notSame(ncbi.taxon('Diphylleia', 'Chloroplastida'),
ott.taxonThatContains('Diphylleia', 'Diphylleia rotans'))
# 2015-10-06 JAR noticed while debugging surprisingly large number of newly-hidden ids
# n.b. there are three Ctenophoras: comb jellies, diatom, fly
ott.notSame(
ncbi.taxon('Ctenophora', 'Metazoa'),
ott.taxon('Ctenophora',
'Stramenopiles')) # diatom from SILVA, WoRMS, GBIF, IRMNG
return ncbi
