# Requires python.security.respectJavaAccessibility = false
# on java command line or in .jython
from org.opentreeoflife.taxa import Taxonomy
from org.opentreeoflife.smasher import UnionTaxonomy, HomonymReport
union = UnionTaxonomy()
skel = Taxonomy.getTaxonomy('tax/skel/', 'skel')
union.setSkeleton(skel)
def report(tax, tag):
union.markDivisionsFromSkeleton(tax, skel)
HomonymReport.homonymReport(tax, 'reports/' + tag + '-homonym-report.tsv')
if True:
ott = Taxonomy.getTaxonomy('tax/ott/', 'ott')
report(ott, 'ott')
else:
import taxonomies
report(taxonomies.loadSilva(), 'silva')
report(taxonomies.loadH2007(), 'h2007')
report(taxonomies.loadFung(), 'worms')
report(taxonomies.loadFung(), 'if')
report(taxonomies.loadNcbi(), 'ncbi')
report(taxonomies.loadGbif(), 'gbif')
report(taxonomies.loadIrmng(), 'irmng')
from org.opentreeoflife.smasher import Taxonomy
from org.opentreeoflife.smasher import Reportx
import taxonomies
ott = Taxonomy.newTaxonomy()
skel = Taxonomy.getTaxonomy('tax/skel/', 'skel')
ott.setSkeleton(skel)
def report(tax, tag):
ott.markDivisions(tax)
# Reportx.bogotypes(tax)
taxonomies.checkDivisions(tax)
Reportx.report(tax, tag + '-mrca-report.tsv')
if True:
report(taxonomies.loadIrmng(), 'irmng')
else:
silva = taxonomies.loadSilva()
ott.notSame(silva.taxon('Ctenophora', 'Coscinodiscophytina'),
skel.taxon('Ctenophora'))
report(silva, 'silva')
report(taxonomies.loadH2007(), 'h2007')
report(taxonomies.loadFung(), 'if')
report(taxonomies.loadNcbi(), 'ncbi')
report(taxonomies.loadGbif(), 'gbif')
report(taxonomies.loadIrmng(), 'irmng')
report(taxonomies.loadOtt(), 'ott')
def doFung():
allfung = taxonomies.loadFung()
print "Index Fungorum has %s nodes"%allfung.count()
# JAR 2014-04-27 JAR found while investigating 'hidden' status of
# Thelohania butleri. Move out of Protozoa to prevent their being hidden
allfung.taxon('Fungi').take(allfung.taxon('Microsporidia'))
# *** Non-Fungi processing
# JAR 2014-05-13 Chlorophyte or fungus? This one is very confused.
# Pick it up from GBIF if at all
allfung.taxon('Byssus phosphorea').prune()
if False: # see taxonomies.loadFung
# Work in progress. By promoting to root we've lost the fact that
# protozoa are eukaryotes, which is unfortunate. Not important in this
# case, but suggestive of deeper problems in the framework.
# Test case: Oomycota should end up in Stramenopiles.
fung_Protozoa = allfung.maybeTaxon('Protozoa')
if fung_Protozoa != None:
fung_Protozoa.hide() # recursive
fung_Protozoa.detach()
fung_Protozoa.elide()
fung_Chromista = allfung.maybeTaxon('Chromista')
if fung_Chromista != None:
fung_Chromista.hide() # recursive
fung_Chromista.detach()
fung_Chromista.elide()
# IF Thraustochytriidae = SILVA Thraustochytriaceae ? (Stramenopiles)
# IF T. 90638 contains Sicyoidochytrium, Schizochytrium, Ulkenia, Thraustochytrium
# Parietichytrium, Elina, Botryochytrium, Althornia
# SILVA T. contains Ulkenia and a few others of these... I say yes.
thraust = allfung.taxon('90377')
thraust.synonym('Thraustochytriaceae')
thraust.synonym('Thraustochytriidae')
thraust.synonym('Thraustochytridae')
# IF Labyrinthulaceae = SILVA Labyrinthulomycetes ? NO.
# IF L. contains only Labyrinthomyxa, Labyrinthula
# SILVA L. contains a lot more than that.
# IF Hyphochytriaceae = SILVA Hyphochytriales ?
# SILVA Hyphochytriales = AB622284/#4 contains only
# Hypochitrium, Rhizidiomycetaceae
# There are two Bacillaria.
# 1. NCBI 3002, in Stramenopiles, contains Bacillaria paxillifer.
# No synonyms in NCBI.
# IF has Bacillaria as a synonym for Camillea (if:777).
# Bacillaria is not otherwise in IF.
# Cammillea in IF is in Stramenopiles.
# 2. NCBI 109369, in Pezizomycotina
# No synonyms in NCBI.
# NCBI 13677 = Camillea, a fish.
# There are two Polyangium, a bacterium (NCBI) and a flatworm (IF).
# smush folds sibling taxa that have the same name.
# (repeats - see loadFung() ???)
allfung.smush()
# *** Fungi processing
print "Fungi in Index Fungorum has %s nodes"%allfung.taxon('Fungi').count()
# fung = allfung.select(allfung.taxon('Fungi'))
# SIDE EFFECT. Ideally there would be a declarative
# (non-side-effecty) way to do this.
# allfung.taxon('Fungi').trim()
# Revert to previous method for the time being due to large chunks of Fungi
# lacking parent links
fung = allfung
# JAR 2014-04-11 Missing in earlier IF, mistake in later IF -
# extraneous authority string. See Romina's issue #42
# This is a fungus.
cyph = fung.maybeTaxon('Cyphellopsis')
if cyph == None:
cyph = fung.maybeTaxon('Cyphellopsis Donk 1931')
if cyph != None:
cyph.rename('Cyphellopsis')
else:
cyph = fung.newTaxon('Cyphellopsis', 'genus', 'if:17439')
fung.taxon('Niaceae').take(cyph)
# 2014-03-07 Prevent a false match
# https://groups.google.com/d/msg/opentreeoflife/5SAPDerun70/fRjA2M6z8tIJ
# This is a fungus in Pezizomycotina
ott.notSame(silva.taxon('Phaeosphaeria'), fung.taxon('Phaeosphaeria'))
# 2014-04-08 This was causing Agaricaceae to be paraphyletic
ott.notSame(silva.taxon('Morganella'), fung.taxon('Morganella'))
# 2014-04-08 More IF/SILVA bad matches
# https://github.com/OpenTreeOfLife/reference-taxonomy/issues/63
for name in ['Trichoderma harzianum', # in Pezizomycotina
'Acantharia', # in Pezizomycotina
'Bogoriella', # in Pezizomycotina
'Steinia', # in Pezizomycotina
'Sclerotinia homoeocarpa', # in Pezizomycotina
'Epiphloea', # in Pezizomycotina
'Campanella', # in Agaricomycotina
'Lacrymaria', # in Agaricomycotina
'Frankia', # in Pezizomycotina / bacterium in SILVA
'Phialina', # in Pezizomycotina
'Puccinia triticina']: # in Pucciniomycotina in Fungi
ott.notSame(silva.taxon(name), fung.taxon(name))
# Romina 2014-04-09
# IF has both Hypocrea and Trichoderma. Hypocrea is the right name.
# https://github.com/OpenTreeOfLife/reference-taxonomy/issues/86
fung.taxon('Trichoderma viride').rename('Hypocrea rufa') # Type
fung.taxon('Hypocrea').absorb(fung.taxonThatContains('Trichoderma', 'Hypocrea rufa'))
# Romina https://github.com/OpenTreeOfLife/reference-taxonomy/issues/42
fung.taxon('Trichoderma deliquescens').rename('Hypocrea lutea')
# 2014-04-25 JAR
# There are three Bostrychias: a rhodophyte, a fungus, and a bird.
# The fungus name is a synonym for Cytospora.
if fung.maybeTaxon('Bostrychia', 'Ascomycota') != None:
ott.notSame(silva.taxon('Bostrychia', 'Rhodophyceae'), fung.taxon('Bostrychia', 'Ascomycota'))
# analyzeMajorRankConflicts sets the "major_rank_conflict" flag when
# intermediate ranks are missing (e.g. a family that's a child of a
# class)
fung.analyzeMajorRankConflicts()
ott.absorb(fung)
print "Fungi in h2007 + if has %s nodes"%ott.taxon('Fungi').count()
# https://github.com/OpenTreeOfLife/reference-taxonomy/issues/20
# Problem: Chlamydotomus is an incertae sedis child of Fungi. Need to
# find a good home for it.
#
# Mycobank says Chlamydotomus beigelii = Trichosporon beigelii:
# http://www.mycobank.org/BioloMICS.aspx?Link=T&TableKey=14682616000000067&Rec=35058&Fields=All
#
# IF says the basionym is Pleurococcus beigelii, and P. beigelii's current name
# is Geotrichum beigelii. IF says the type for Trichosporon is Trichosporon beigelii,
# and that T. beigelii's current name is Trichosporum beigelii... with no synonymies...
# So IF does not corroborate Mycobank.
#
# So we could consider absorbing Chlamydotomus into Trichosoporon. But...
#
# Not sure about this. beigelii has a sister, cellaris, that should move along
# with it, but the name Trichosporon cellaris has never been published.
# Cb = ott.taxon('Chlamydotomus beigelii')
# Cb.rename('Trichosporon beigelii')
# ott.taxon('Trichosporon').take(Cb)
#
# Just make it incertae sedis and put off dealing with it until someone cares...
# 2014-04-13 Romina #40, #60
for foo in [('Neozygitales', ['Neozygitaceae']),
('Asteriniales', ['Asterinaceae']),
('Savoryellales', ['Savoryella', 'Ascotaiwania', 'Ascothailandia']),
('Cladochytriales', ['Cladochytriaceae', 'Nowakowskiellaceae', 'Septochytriaceae', 'Endochytriaceae']),
('Jaapiales', ['Jaapiaceae']),
('Coniocybales', ['Coniocybaceae']),
('Hyaloraphidiales', ['Hyaloraphidiaceae']),
('Mytilinidiales', ['Mytilinidiaceae', 'Gloniaceae'])]:
order = ott.taxon(foo[0])
for family in foo[1]:
order.take(ott.taxon(family))
# ** No taxon found with this name: Nowakowskiellaceae
# ** No taxon found with this name: Septochytriaceae
# ** No taxon found with this name: Jaapiaceae
# ** (null=if:81865 Rhizocarpaceae) is already a child of (null=h2007:212 Rhizocarpales)
# ** No taxon found with this name: Hyaloraphidiaceae
return fung
def doFung():
allfung = taxonomies.loadFung()
print "Index Fungorum has %s nodes"%allfung.count()
# JAR 2014-04-27 JAR found while investigating 'hidden' status of
# Thelohania butleri. Move out of Protozoa to prevent their being hidden
allfung.taxon('Fungi').take(allfung.taxon('Microsporidia'))
# *** Non-Fungi processing
# JAR 2014-05-13 Chlorophyte or fungus? This one is very confused.
# Pick it up from GBIF if at all
allfung.taxon('Byssus phosphorea').prune()
if False: # see taxonomies.loadFung
# Work in progress. By promoting to root we've lost the fact that
# protozoa are eukaryotes, which is unfortunate. Not important in this
# case, but suggestive of deeper problems in the framework.
# Test case: Oomycota should end up in Stramenopiles.
fung_Protozoa = allfung.maybeTaxon('Protozoa')
if fung_Protozoa != None:
fung_Protozoa.hide() # recursive
fung_Protozoa.detach()
fung_Protozoa.elide()
fung_Chromista = allfung.maybeTaxon('Chromista')
if fung_Chromista != None:
fung_Chromista.hide() # recursive
fung_Chromista.detach()
fung_Chromista.elide()
# IF Thraustochytriidae = SILVA Thraustochytriaceae ? (Stramenopiles)
# IF T. 90638 contains Sicyoidochytrium, Schizochytrium, Ulkenia, Thraustochytrium
# Parietichytrium, Elina, Botryochytrium, Althornia
# SILVA T. contains Ulkenia and a few others of these... I say yes.
thraust = allfung.taxon('90377')
thraust.synonym('Thraustochytriaceae')
thraust.synonym('Thraustochytriidae')
thraust.synonym('Thraustochytridae')
# IF Labyrinthulaceae = SILVA Labyrinthulomycetes ? NO.
# IF L. contains only Labyrinthomyxa, Labyrinthula
# SILVA L. contains a lot more than that.
# IF Hyphochytriaceae = SILVA Hyphochytriales ?
# SILVA Hyphochytriales = AB622284/#4 contains only
# Hypochitrium, Rhizidiomycetaceae
# There are two Bacillaria.
# 1. NCBI 3002, in Stramenopiles, contains Bacillaria paxillifer.
# No synonyms in NCBI.
# IF has Bacillaria as a synonym for Camillea (if:777).
# Bacillaria is not otherwise in IF.
# Cammillea in IF is in Stramenopiles.
# 2. NCBI 109369, in Pezizomycotina
# No synonyms in NCBI.
# NCBI 13677 = Camillea, a fish.
# There are two Polyangium, a bacterium (NCBI) and a flatworm (IF).
# smush folds sibling taxa that have the same name.
# (repeats - see loadFung() ???)
allfung.smush()
# *** Fungi processing
print "Fungi in Index Fungorum has %s nodes"%allfung.taxon('Fungi').count()
# fung = allfung.select(allfung.taxon('Fungi'))
# SIDE EFFECT. Ideally there would be a declarative
# (non-side-effecty) way to do this.
# allfung.taxon('Fungi').trim()
# Revert to previous method for the time being due to large chunks of Fungi
# lacking parent links
fung = allfung
# JAR 2014-04-11 Missing in earlier IF, mistake in later IF -
# extraneous authority string. See Romina's issue #42
# This is a fungus.
cyph = fung.maybeTaxon('Cyphellopsis')
if cyph == None:
cyph = fung.maybeTaxon('Cyphellopsis Donk 1931')
if cyph != None:
cyph.rename('Cyphellopsis')
else:
cyph = fung.newTaxon('Cyphellopsis', 'genus', 'if:17439')
fung.taxon('Niaceae').take(cyph)
# 2014-03-07 Prevent a false match
# https://groups.google.com/d/msg/opentreeoflife/5SAPDerun70/fRjA2M6z8tIJ
# This is a fungus in Pezizomycotina
ott.notSame(silva.taxon('Phaeosphaeria'), fung.taxon('Phaeosphaeria'))
# 2014-04-08 This was causing Agaricaceae to be paraphyletic
ott.notSame(silva.taxon('Morganella'), fung.taxon('Morganella'))
# 2014-04-08 More IF/SILVA bad matches
# https://github.com/OpenTreeOfLife/reference-taxonomy/issues/63
for name in ['Trichoderma harzianum', # in Pezizomycotina
'Acantharia', # in Pezizomycotina
'Bogoriella', # in Pezizomycotina
'Steinia', # in Pezizomycotina
'Sclerotinia homoeocarpa', # in Pezizomycotina
'Epiphloea', # in Pezizomycotina
'Campanella', # in Agaricomycotina
'Lacrymaria', # in Agaricomycotina
'Frankia', # in Pezizomycotina / bacterium in SILVA
'Phialina', # in Pezizomycotina
'Puccinia triticina']: # in Pucciniomycotina in Fungi
ott.notSame(silva.taxon(name), fung.taxon(name))
# Romina 2014-04-09
# IF has both Hypocrea and Trichoderma. Hypocrea is the right name.
# https://github.com/OpenTreeOfLife/reference-taxonomy/issues/86
fung.taxon('Trichoderma viride').rename('Hypocrea rufa') # Type
fung.taxon('Hypocrea').absorb(fung.taxonThatContains('Trichoderma', 'Hypocrea rufa'))
# Romina https://github.com/OpenTreeOfLife/reference-taxonomy/issues/42
fung.taxon('Trichoderma deliquescens').rename('Hypocrea lutea')
# 2014-04-25 JAR
# There are three Bostrychias: a rhodophyte, a fungus, and a bird.
# The fungus name is a synonym for Cytospora.
if fung.maybeTaxon('Bostrychia', 'Ascomycota') != None:
ott.notSame(silva.taxon('Bostrychia', 'Rhodophyceae'), fung.taxon('Bostrychia', 'Ascomycota'))
# analyzeMajorRankConflicts sets the "major_rank_conflict" flag when
# intermediate ranks are missing (e.g. a family that's a child of a
# class)
fung.analyzeMajorRankConflicts()
ott.absorb(fung)
print "Fungi in h2007 + if has %s nodes"%ott.taxon('Fungi').count()
# https://github.com/OpenTreeOfLife/reference-taxonomy/issues/20
# Problem: Chlamydotomus is an incertae sedis child of Fungi. Need to
# find a good home for it.
#
# Mycobank says Chlamydotomus beigelii = Trichosporon beigelii:
# http://www.mycobank.org/BioloMICS.aspx?Link=T&TableKey=14682616000000067&Rec=35058&Fields=All
#
# IF says the basionym is Pleurococcus beigelii, and P. beigelii's current name
# is Geotrichum beigelii. IF says the type for Trichosporon is Trichosporon beigelii,
# and that T. beigelii's current name is Trichosporum beigelii... with no synonymies...
# So IF does not corroborate Mycobank.
#
# So we could consider absorbing Chlamydotomus into Trichosoporon. But...
#
# Not sure about this. beigelii has a sister, cellaris, that should move along
# with it, but the name Trichosporon cellaris has never been published.
# Cb = ott.taxon('Chlamydotomus beigelii')
# Cb.rename('Trichosporon beigelii')
# ott.taxon('Trichosporon').take(Cb)
#
# Just make it incertae sedis and put off dealing with it until someone cares...
# 2014-04-13 Romina #40, #60
for foo in [('Neozygitales', ['Neozygitaceae']),
('Asteriniales', ['Asterinaceae']),
('Savoryellales', ['Savoryella', 'Ascotaiwania', 'Ascothailandia']),
('Cladochytriales', ['Cladochytriaceae', 'Nowakowskiellaceae', 'Septochytriaceae', 'Endochytriaceae']),
('Jaapiales', ['Jaapiaceae']),
('Coniocybales', ['Coniocybaceae']),
('Hyaloraphidiales', ['Hyaloraphidiaceae']),
('Mytilinidiales', ['Mytilinidiaceae', 'Gloniaceae'])]:
order = ott.taxon(foo[0])
for family in foo[1]:
order.take(ott.taxon(family))
# ** No taxon found with this name: Nowakowskiellaceae
# ** No taxon found with this name: Septochytriaceae
# ** No taxon found with this name: Jaapiaceae
# ** (null=if:81865 Rhizocarpaceae) is already a child of (null=h2007:212 Rhizocarpales)
# ** No taxon found with this name: Hyaloraphidiaceae
return fung
# Requires python.security.respectJavaAccessibility = false
# on java command line or in .jython
from org.opentreeoflife.taxa import Taxonomy
from org.opentreeoflife.smasher import UnionTaxonomy, HomonymReport
union = UnionTaxonomy()
skel = Taxonomy.getTaxonomy('tax/skel/', 'skel')
union.setSkeleton(skel)
def report(tax, tag):
union.markDivisionsFromSkeleton(tax, skel)
HomonymReport.homonymReport(tax, 'reports/' + tag + '-homonym-report.tsv')
if True:
ott = Taxonomy.getTaxonomy('tax/ott/', 'ott')
report(ott, 'ott')
else:
import taxonomies
report(taxonomies.loadSilva(), 'silva')
report(taxonomies.loadH2007(), 'h2007')
report(taxonomies.loadFung(), 'worms')
report(taxonomies.loadFung(), 'if')
report(taxonomies.loadNcbi(), 'ncbi')
report(taxonomies.loadGbif(), 'gbif')
report(taxonomies.loadIrmng(), 'irmng')
from org.opentreeoflife.smasher import Taxonomy
from org.opentreeoflife.smasher import Reportx
import taxonomies
ott = Taxonomy.newTaxonomy()
skel = Taxonomy.getTaxonomy('tax/skel/', 'skel')
ott.setSkeleton(skel)
def report(tax, tag):
ott.markDivisions(tax)
# Reportx.bogotypes(tax)
taxonomies.checkDivisions(tax)
Reportx.report(tax, tag + '-mrca-report.tsv')
if True:
report(taxonomies.loadIrmng(), 'irmng')
else:
silva = taxonomies.loadSilva()
ott.notSame(silva.taxon('Ctenophora', 'Coscinodiscophytina'),
skel.taxon('Ctenophora'))
report(silva, 'silva')
report(taxonomies.loadH2007(), 'h2007')
report(taxonomies.loadFung(), 'if')
report(taxonomies.loadNcbi(), 'ncbi')
report(taxonomies.loadGbif(), 'gbif')
report(taxonomies.loadIrmng(), 'irmng')
report(taxonomies.loadOtt(), 'ott')
