def buildCodonML( codon_model = "f3x4-four",
grammar_type = "linear",
num_blocks = 2,
explicit_extension = False,
fix_omega = False,
fix_kappa = False,
fix_frequencies = False,
shared_frequencies = False,
shared_rates = False,
annotate_terminals = None,
codon_frequencies = None ):
"""build a codonml-like model.
fix_omega: If true, omega is set to 1. Thus the model does not distinguish
between synonymous and non-synonymous transitions. Use this option
to estimate the number of synonymous and non-synonymous sites.
(The original rate parameters are per codon.)
fix_kappa: If true, kappa is set to 1. Thus the model does not estimate
the transition/transversion ratio.
fix_frequencies: if True: frequencies of the codons are set to const
and are not estimated. The corresponding rate parameters should be
added afterwards.
grammar_type can be one of "linear", "linear-blocks".
linear: simple linear sequence with gaps
linear-blocks: blocks of linear sequence
codon_model can be one of "f3x4-two", "f3x4-four"
f3x4-two: two parameter parameterization (a al CodonML)
Codon frequencies are given by the nucleotide frequencies at each position.
f3x4-four: four parameter parameterization
(better convergence properties).
Codon frequencies are given by the nucleotide frequencies at each position.
codons-four: four parameter parametrization with explicitely
given codon frequencies. These are added as constant
expressions into the grammar.
num_blocks: the number of blocks for the "linear-blocks" grammar.
shared_frequencies: the nucleotide frequencies are shared between blocks.
The default (false) has separate nucleotide frequencies for each block.
shared_rates: the rates between blocks are shared. The default (false)
has separate rates for each block.
annotate_terminals:
The linear-blocks grammar allows to annotate the input
alignments. Provide a hash mapping the emitted states to
Annotations.
explicit_extension:
use an explicit parameter for the markov chain.
"""
model = XGram.Model.Model()
grammar = XGram.Model.Grammar()
alphabet = XGram.Model.Alphabet()
if grammar_type == "linear":
chain = XGram.Model.Chain()
generator_chain = ChainCodonML(fix_omega = fix_omega,
fix_kappa = fix_kappa,
fix_frequencies = fix_frequencies,
codon_frequencies = codon_frequencies,
codon_model = codon_model )
generator_chain.buildGrammar( chain )
grammar.addChain( chain )
generator_grammar = GrammarLinearSequence( explicit_extension = explicit_extension )
elif grammar_type == "linear-blocks":
for x in range( num_blocks ):
chain = XGram.Model.Chain()
generator_chain = ChainCodonML(fix_omega = fix_omega,
fix_kappa = fix_kappa,
fix_frequencies = fix_frequencies,
shared_frequencies = shared_frequencies,
shared_rates = shared_rates,
codon_model = codon_model,
codon_frequencies = codon_frequencies,
prefix = "B%x_" % x )
generator_chain.buildGrammar( chain )
grammar.addChain( chain )
generator_grammar = GrammarLinearSequenceBlocks( annotate_terminals = annotate_terminals )
else:
raise XGram.Exceptions.UsageError( 'unknown grammar %s' % grammar_type )
generator_grammar.buildGrammar( grammar )
grammar.setName( "codonML" )
generator_alphabet = AlphabetDNA()
generator_alphabet.buildGrammar(alphabet)
model.setAlphabet( alphabet )
model.setGrammar( grammar )
return model
def buildModel( substitution_model = "jc69",
grammar_type = "linear",
num_blocks = 1,
shared_frequencies = False,
shared_rates = False,
annotate_terminals = False,
explicit_extension = False,
):
"""build a standard nucleotide models.
model: can be either one of
jc69: Jukes-Cantor model
k80: Kimura 2-parameter model
gtr: general time reversible model (9 parameters)
rev: general 12 parameter model
exclicit_extension: add an explicit extension probability
(called ext and not_ext) to the grammar.
grammar_type can be one of "linear", "linear-blocks".
linear: simple linear sequence with gaps
linear-blocks: blocks of linear sequence
num_blocks: the number of blocks for the "linear-blocks" grammar.
shared_frequencies: the nucleotide frequencies are shared between blocks.
The default (false) has separate nucleotide frequencies for each block.
shared_rates: the rates between blocks are shared. The default (false)
has separate rates for each block.
annotate_terminals:
The linear-blocks grammar allows to annotate the input
alignments. Provide a hash mapping the emitted states to
Annotations.
"""
model = XGram.Model.Model()
grammar = XGram.Model.Grammar()
alphabet = XGram.Model.Alphabet()
if substitution_model == "jc69":
generator_initial_states = InitialStatesConst
generator_transitions = TransitionsConst
elif substitution_model == "k80":
generator_initial_states = InitialStatesConst
generator_transitions = TransitionsK80
elif substitution_model == "rev":
generator_initial_states = InitialStatesConst
generator_transitions = TransitionsConst
elif substitution_model == "gtr":
generator_initial_states = InitialStatesParametric
generator_transitions = TransitionsGTR
else:
raise Exceptions.UsageError( "model %s unknown" % substitution_model )
generator_grammar = GrammarLinearSequence( explicit_extension = explicit_extension )
if grammar_type == "linear":
chain = XGram.Model.Chain()
generator_chain = ChainDNA( generator_initial_states(),
generator_transitions() )
generator_chain.buildGrammar( chain )
if substitution_model == "rev":
chain.setUpdatePolicy( "rev" )
grammar.addChain( chain )
generator_grammar = GrammarLinearSequence( explicit_extension = explicit_extension )
elif grammar_type in ("linear-blocks", "multiple-blocks" ) :
for x in range( num_blocks ):
chain = XGram.Model.Chain()
prefix = "B%x" % x
generator_chain = ChainDNA( generator_initial_states( shared_rates = shared_rates,
shared_frequencies = shared_frequencies,
prefix = prefix ),
generator_transitions( shared_rates = shared_rates,
shared_frequencies = shared_frequencies,
prefix=prefix ),
prefix = prefix )
if substitution_model == "rev":
chain.setUpdatePolicy( "rev" )
generator_chain.buildGrammar( chain )
grammar.addChain( chain )
if grammar_type == "multiple-blocks":
generator_grammar = GrammarLinearSequenceMultipleChains( annotate_terminals = annotate_terminals )
elif grammar_type == "linear-blocks":
generator_grammar = GrammarLinearSequenceBlocks( annotate_terminals = annotate_terminals )
else:
raise XGram.Exceptions.UsageError( 'unknown grammar %s' % grammar_type )
generator_grammar.buildGrammar( grammar )
grammar.setName( substitution_model )
grammar.addComment( "grammar built by DNA.py " )
grammar.addComment( " shared rates = %s" % shared_rates)
grammar.addComment( " shared frequencies = %s" % shared_frequencies)
grammar.addComment( " type = %s" % grammar_type)
grammar.addComment( " nblocks = %i" % num_blocks)
grammar.addComment( " substitution model = %s" % substitution_model)
generator_alphabet = AlphabetDNA()
generator_alphabet.buildGrammar(alphabet)
model.setAlphabet( alphabet )
model.setGrammar( grammar )
return model
def buildCodonML(codon_model="f3x4-four",
grammar_type="linear",
num_blocks=2,
explicit_extension=False,
fix_omega=False,
fix_kappa=False,
fix_frequencies=False,
shared_frequencies=False,
shared_rates=False,
annotate_terminals=None,
codon_frequencies=None):
"""build a codonml-like model.
fix_omega: If true, omega is set to 1. Thus the model does not distinguish
between synonymous and non-synonymous transitions. Use this option
to estimate the number of synonymous and non-synonymous sites.
(The original rate parameters are per codon.)
fix_kappa: If true, kappa is set to 1. Thus the model does not estimate
the transition/transversion ratio.
fix_frequencies: if True: frequencies of the codons are set to const
and are not estimated. The corresponding rate parameters should be
added afterwards.
grammar_type can be one of "linear", "linear-blocks".
linear: simple linear sequence with gaps
linear-blocks: blocks of linear sequence
codon_model can be one of "f3x4-two", "f3x4-four"
f3x4-two: two parameter parameterization (a al CodonML)
Codon frequencies are given by the nucleotide frequencies at each position.
f3x4-four: four parameter parameterization
(better convergence properties).
Codon frequencies are given by the nucleotide frequencies at each position.
codons-four: four parameter parametrization with explicitely
given codon frequencies. These are added as constant
expressions into the grammar.
num_blocks: the number of blocks for the "linear-blocks" grammar.
shared_frequencies: the nucleotide frequencies are shared between blocks.
The default (false) has separate nucleotide frequencies for each block.
shared_rates: the rates between blocks are shared. The default (false)
has separate rates for each block.
annotate_terminals:
The linear-blocks grammar allows to annotate the input
alignments. Provide a hash mapping the emitted states to
Annotations.
explicit_extension:
use an explicit parameter for the markov chain.
"""
model = XGram.Model.Model()
grammar = XGram.Model.Grammar()
alphabet = XGram.Model.Alphabet()
if grammar_type == "linear":
chain = XGram.Model.Chain()
generator_chain = ChainCodonML(fix_omega=fix_omega,
fix_kappa=fix_kappa,
fix_frequencies=fix_frequencies,
codon_frequencies=codon_frequencies,
codon_model=codon_model)
generator_chain.buildGrammar(chain)
grammar.addChain(chain)
generator_grammar = GrammarLinearSequence(
explicit_extension=explicit_extension)
elif grammar_type == "linear-blocks":
for x in range(num_blocks):
chain = XGram.Model.Chain()
generator_chain = ChainCodonML(
fix_omega=fix_omega,
fix_kappa=fix_kappa,
fix_frequencies=fix_frequencies,
shared_frequencies=shared_frequencies,
shared_rates=shared_rates,
codon_model=codon_model,
codon_frequencies=codon_frequencies,
prefix="B%x_" % x)
generator_chain.buildGrammar(chain)
grammar.addChain(chain)
generator_grammar = GrammarLinearSequenceBlocks(
annotate_terminals=annotate_terminals)
else:
raise XGram.Exceptions.UsageError('unknown grammar %s' % grammar_type)
generator_grammar.buildGrammar(grammar)
grammar.setName("codonML")
generator_alphabet = AlphabetDNA()
generator_alphabet.buildGrammar(alphabet)
model.setAlphabet(alphabet)
model.setGrammar(grammar)
return model
def buildModel(
substitution_model="jc69",
grammar_type="linear",
num_blocks=1,
shared_frequencies=False,
shared_rates=False,
annotate_terminals=False,
explicit_extension=False,
):
"""build a standard nucleotide models.
model: can be either one of
jc69: Jukes-Cantor model
k80: Kimura 2-parameter model
gtr: general time reversible model (9 parameters)
rev: general 12 parameter model
exclicit_extension: add an explicit extension probability
(called ext and not_ext) to the grammar.
grammar_type can be one of "linear", "linear-blocks".
linear: simple linear sequence with gaps
linear-blocks: blocks of linear sequence
num_blocks: the number of blocks for the "linear-blocks" grammar.
shared_frequencies: the nucleotide frequencies are shared between blocks.
The default (false) has separate nucleotide frequencies for each block.
shared_rates: the rates between blocks are shared. The default (false)
has separate rates for each block.
annotate_terminals:
The linear-blocks grammar allows to annotate the input
alignments. Provide a hash mapping the emitted states to
Annotations.
"""
model = XGram.Model.Model()
grammar = XGram.Model.Grammar()
alphabet = XGram.Model.Alphabet()
if substitution_model == "jc69":
generator_initial_states = InitialStatesConst
generator_transitions = TransitionsConst
elif substitution_model == "k80":
generator_initial_states = InitialStatesConst
generator_transitions = TransitionsK80
elif substitution_model == "rev":
generator_initial_states = InitialStatesConst
generator_transitions = TransitionsConst
elif substitution_model == "gtr":
generator_initial_states = InitialStatesParametric
generator_transitions = TransitionsGTR
else:
raise Exceptions.UsageError("model %s unknown" % substitution_model)
generator_grammar = GrammarLinearSequence(
explicit_extension=explicit_extension)
if grammar_type == "linear":
chain = XGram.Model.Chain()
generator_chain = ChainDNA(generator_initial_states(),
generator_transitions())
generator_chain.buildGrammar(chain)
if substitution_model == "rev":
chain.setUpdatePolicy("rev")
grammar.addChain(chain)
generator_grammar = GrammarLinearSequence(
explicit_extension=explicit_extension)
elif grammar_type in ("linear-blocks", "multiple-blocks"):
for x in range(num_blocks):
chain = XGram.Model.Chain()
prefix = "B%x" % x
generator_chain = ChainDNA(
generator_initial_states(shared_rates=shared_rates,
shared_frequencies=shared_frequencies,
prefix=prefix),
generator_transitions(shared_rates=shared_rates,
shared_frequencies=shared_frequencies,
prefix=prefix),
prefix=prefix)
if substitution_model == "rev":
chain.setUpdatePolicy("rev")
generator_chain.buildGrammar(chain)
grammar.addChain(chain)
if grammar_type == "multiple-blocks":
generator_grammar = GrammarLinearSequenceMultipleChains(
annotate_terminals=annotate_terminals)
elif grammar_type == "linear-blocks":
generator_grammar = GrammarLinearSequenceBlocks(
annotate_terminals=annotate_terminals)
else:
raise XGram.Exceptions.UsageError('unknown grammar %s' % grammar_type)
generator_grammar.buildGrammar(grammar)
grammar.setName(substitution_model)
grammar.addComment("grammar built by DNA.py ")
grammar.addComment(" shared rates = %s" % shared_rates)
grammar.addComment(" shared frequencies = %s" % shared_frequencies)
grammar.addComment(" type = %s" % grammar_type)
grammar.addComment(" nblocks = %i" % num_blocks)
grammar.addComment(" substitution model = %s" % substitution_model)
generator_alphabet = AlphabetDNA()
generator_alphabet.buildGrammar(alphabet)
model.setAlphabet(alphabet)
model.setGrammar(grammar)
return model