#!/usr/bin/env python
"""Classes for dealing with base, codon, and amino acid usage.
"""
from __future__ import division
from cogent.maths.stats.util import Freqs, Numbers, UnsafeFreqs
from cogent.maths.stats.special import fix_rounding_error
from cogent.util.array import euclidean_distance
from cogent.util.misc import Delegator, FunctionWrapper, InverseDict
from cogent.core.genetic_code import GeneticCodes, GeneticCode as GenCodeClass
from cogent.core.info import Info as InfoClass
from cogent.core.alphabet import CharAlphabet
from string import upper
from numpy import array, concatenate, sum, mean, isfinite, sqrt
__author__ = "Rob Knight"
__copyright__ = "Copyright 2007-2012, The Cogent Project"
__credits__ = ["Rob Knight", "Sandra Smit", "Daniel McDonald"]
__license__ = "GPL"
__version__ = "1.5.3"
__maintainer__ = "Rob Knight"
__email__ = "rob@spot.colorado.edu"
__status__ = "Production"
RnaBases = CharAlphabet('UCAG')
DnaBases = CharAlphabet('TCAG')
AminoAcids = CharAlphabet('ACDEFGHIKLMNPQRSTVWY*') # * denotes termination
AB = CharAlphabet('ab') #used for testing
Chars = CharAlphabet(''.join(map(chr, range(256))), '-') #used for raw chars
RnaBasesGap = CharAlphabet('UCAG-', '-')
DnaBasesGap = CharAlphabet('TCAG-', '-')
AminoAcidsGap = CharAlphabet('ACDEFGHIKLMNPQRSTVWY*-', '-')
DnaIupac = CharAlphabet('TCAGNVBHDKSWMYR')
RnaIupac = CharAlphabet('UCAGNVBHDKSWMYR')
AminoAcidsIupac = CharAlphabet('ACDEFGHIKLMNPQRSTVWY*XBZ')
DnaIupacGap = CharAlphabet('TCAG-NVBHDKSWMYR', '-')
RnaIupacGap = CharAlphabet('UCAG-NVBHDKSWMYR', '-')
AminoAcidsIupacGap = CharAlphabet('ACDEFGHIKLMNPQRSTVWY*-XBZ', '-')
RnaPairs = RnaBases**2
DnaPairs = DnaBases**2
RnaGapPairs = RnaBasesGap**2
DnaGapPairs = DnaBasesGap**2
AminoAcidPairs = AminoAcids**2
ABPairs = AB**2
#RnaBases = 'UCAG'
#DnaBases = 'TCAG'
RnaCodons = [i+j+k for i in RnaBases for j in RnaBases for k in RnaBases]
DnaCodons = [i+j+k for i in DnaBases for j in DnaBases for k in DnaBases]
#AminoAcids = 'ACDEFGHIKLMNPQRSTVWY*'
SGC = GeneticCodes[1]
RnaDinucs = [i+j for i in RnaBases for j in RnaBases]
RnaToDna = dict(zip(RnaBases, DnaBases))
DnaToRna = dict(zip(DnaBases, RnaBases))
Bases = RnaBases #by default
Codons = RnaCodons #by default
_equal_bases = Freqs(Bases)
_equal_codons = Freqs(Codons)
_equal_amino_acids = Freqs(AminoAcids[:-1]) #exclude Stop
for i in (_equal_bases, _equal_codons, _equal_amino_acids):
i.normalize()
empty_rna_codons = dict.fromkeys(RnaCodons, 0.0)
empty_dna_codons = dict.fromkeys(DnaCodons, 0.0)
def seq_to_codon_dict(seq, empty_codons=empty_dna_codons):
"""Converts sequence into codon dict."""
leftover = len(seq) % 3
if leftover:
seq += 'A' * (3-leftover)
result = empty_codons.copy()
for i in range(0, len(seq), 3):
curr = seq[i:i+3]
if curr in result: #ignore others
result[curr] += 1
return result
def UnsafeCodonsFromString(seq, rna=False, formatted=False, **kwargs):
if rna:
d = empty_rna_codons
if not formatted:
seq = seq.upper().replace('T','U')
else:
d = empty_dna_codons
if not formatted:
seq = seq.upper().replace('U','T')
return UnsafeCodonUsage(seq_to_codon_dict(seq,d), **kwargs)
def key_to_rna(key):
"""Sets key to uppercase RNA."""
return key.upper().replace('T', 'U')
def key_to_dna(key):
"""Sets key to uppercase DNA."""
return key.upper().replace('U', 'T')
class InfoFreqs(Freqs, Delegator):
"""Like Freqs, but has an Info object storing additional data.
Intended for holding base or codon frequencies that come from a
particular sequence, so that the Info of the sequence can be preserved
even if the sequence is deleted to free up memory.
"""
def __init__(self, data=None, Info=None, **kwargs):
"""Intializes BaseUsage with data, either sequence or dict of freqs.
Ignores additional kwargs (e.g. to support copy).
Makes the _handler for delegator accessible with the name Info.
"""
if Info is None:
if hasattr(data, 'Info'):
Info = data.Info
else:
Info = InfoClass()
Delegator.__init__(self, Info)
Freqs.__init__(self, data or [], **kwargs)
def _get_info(self):
"""Accessor for Info."""
return self._handler
def _set_info(self, obj):
"""Mutator for Info."""
self._handler = obj
Info = property(_get_info, _set_info)
class BaseUsageI(object):
"""Provides shared interface for BaseUsage classes.
BaseUsage stores counts of the four DNA or RNA bases.
"""
def bases(self):
"""Supports bases/codons/positionalBases/aminoAcids interface."""
return self
def codons(self):
"""Predicts codon frequencies from the base frequencies."""
result = {}
base_copy = self.__class__(self)
base_copy.normalize()
for c in Codons:
curr = 1
for i in c:
curr *= base_copy[i]
result[c] = curr
return CodonUsage(result, self.Info)
def positionalBases(self):
"""Returns PositionalBaseUsage with copy of self at each position."""
return PositionalBaseUsage(self.__class__(self), self.__class__(self),
self.__class__(self), self.Info)
def aminoAcids(self, genetic_code=SGC):
"""Predicts amino acid frequencies from the base frequencies."""
return self.codons().aminoAcids(genetic_code)
def distance(self,other):
"""Calculates the distance between two BaseUsages.
Distance is measured in three directions, CG-content, CU-content, and
GU-content.
"""
return euclidean_distance(array(self.toCartesian()),\
array(other.toCartesian()))
def content(self, string):
"""Gets the sum of bases specified in string.
For example, self.content('GC') gets the GC content.
"""
return sum([self.get(i, 0) for i in string], 0)
def toCartesian(self):
"""Returns tuple of x, y, z coordinates from BaseUsage.
x=u+c, y=u+g, z=u+a
Doesn't alter original object.
"""
return self['UC'], self['UG'], self['UA']
def fromCartesian(cls, *coords):
"""Returns new BaseUsage with A,C,G,U coordinates from UC,UG,UA.
From UC,UG,UA to A,C,G(,U).
This will only work when the original coordinates come from a simplex,
where U+C+A+G=1
"""
result = cls()
x,y,z = coords
u = fix_rounding_error((1-x-y-z)/-2)
a, c, g = z-u, x-u, y-u
result['A'] = a
result['C'] = c
result['G'] = g
result['U'] = u
return result
fromCartesian = classmethod(fromCartesian)
class BaseUsage(BaseUsageI, InfoFreqs):
"""Stores frequencies of the four bases, mapped to RNA.
This class is convenient but inefficient, since it automatically maps any
lookups to the uppercase RNA alphabet internally. Use UnsafeBaseUsage for
speed when necessary.
"""
Mask = FunctionWrapper(key_to_rna)
RequiredKeys = dict.fromkeys(Bases)
def __getitem__(self, key):
"""Normalizes key and treats T=U."""
key = self.Mask(key)
if len(key) == 2: #pair of bases, e.g. GC for GC content
dup = BaseUsage(self)
dup.normalize()
return sum([dup.get(i,0) for i in key], 0)
else:
return super(BaseUsage, self).__getitem__(key)
class UnsafeBaseUsage(BaseUsageI, UnsafeFreqs):
"""Stores frequencies of the four bases. Does not do any validation.
This class avoids overriding most built-ins, so is much faster than
BaseFreqs (although it is often less convenient).
"""
RequiredKeys = dict.fromkeys(Bases)
Info = {} # for interface compatibility with InfoFreqs-based class
class CodonUsageI(object):
"""Stores codon usage for a gene or species.
Note that CodonUsage objects get their own reference to _default_code
during creation, so changing CodonUsage._default_code will not change the
GeneticCode of any CodonUsage object that has already been created.
"""
_default_code = SGC
BlockAbbreviations = \
{'UU':'F/L', 'CU':'Leu', 'AU':'I/M', 'GU':'Val', \
'UC':'Ser', 'CC':'Pro', 'AC':'Thr', 'GC':'Ala',\
'UA':'Y/*','CA':'H/Q','AA':'N/K','GA':'D/E',\
'UG':'C*W', 'CG':'Arg', 'AG':'S/R', 'GG':'Gly'}
BlockNames = \
{'UU':'Phe/Leu', 'CU':'Leucine', 'AU':'Ile/Met', 'GU':'Valine', \
'UC':'Serine', 'CC':'Proline', 'AC':'Threonine', 'GC':'Alanine',\
'UA':'Tyr/Ter','CA':'His/Gln','AA':'Asn/Lys','GA':'Asp/Glu',\
'UG':'Cys/Ter/Trp', 'CG':'Arginine', 'AG':'Ser/Arg', 'GG':'Glycine'}
Blocks = [i+j for i in 'UCAG' for j in 'UCAG'] #UCAG order
SingleAABlocks = ['GC','CG','GG','CU','CC','UC','AC','GU'] #alpha by aa
SplitBlocks = ['UU', 'CA','AU','AA','AG','GA'] #UCAG order]
AbbreviationsToBlocks = InverseDict(BlockAbbreviations)
NamesToBlocks = InverseDict(BlockNames)
BaseUsageClass = None #Must overrride in subclasses
def bases(self, purge_unwanted=False):
"""Returns overall base counts."""
result = {}
if purge_unwanted:
data = self._purged_data()
else:
data = self
for codon, freq in data.items():
for base in codon:
if base in result:
result[base] += freq
else:
result[base] = freq
return self.BaseUsageClass(result, self.Info)
def codons(self):
"""Supports codons/aminoAcids/bases/positionalBases interface."""
return self
def rscu(self):
"""Normalizes self in-place to RSCU, relative synonymous codon usage.
RSCU divides the frequency of each codon to the sum of the freqs for
that codon's amino acid.
"""
gc = self.GeneticCode
syn = gc.Synonyms
aa_sums = {}
for key, codons in syn.items():
aa_sums[key] = sum([self[c] for c in codons], 0)
for codon in self:
try:
curr = self[codon]
res = curr/aa_sums[gc[codon]]
except (KeyError, ZeroDivisionError, FloatingPointError):
pass
else:
if isfinite(res):
self[codon] = res
return self
def _purged_data(self):
"""Copy of self's freqs after removing bad/stop codons and singlets."""
good_codons = self.RequiredKeys
code = self.GeneticCode
data = dict(self) #need copy, since we're deleting required keys
#delete any bad codons
for codon in self:
if codon not in good_codons:
del data[codon]
#delete any stop codons in the current code
for codon in code['*']:
try:
c = codon.upper().replace('T','U')
del data[c]
except KeyError:
pass #don't care if it's not there
#delete any single-item blocks in the current code (i.e. leaving
#only doublets and quartets).
for group in code.Blocks:
if len(group) == 1:
try:
c = group[0].upper().replace('T','U')
del data[c]
except KeyError:
pass #don't care if already deleted
return data
def positionalBases(self, purge_unwanted=False):
"""Calculates positional base usage.
purge_unwanted controls whether or not to purge 1-codon groups, stop
codons, and any codons containing degnerate bases before calculating
the base usage (e.g. to get Noboru Sueoka's P3 measurement): default
is False. Deletion of unwanted codons happens on a copy, not the
original data.
"""
first = {}
second = {}
third = {}
if purge_unwanted: #make a copy of the data and delete things from it
data = self._purged_data()
else:
data = self
for codon, freq in data.items():
try:
p1, p2, p3 = codon
except ValueError:
continue #skip any incomplete codons
if p1 in first:
first[p1] += freq
else:
first[p1] = freq
if p2 in second:
second[p2] += freq
else:
second[p2] = freq
if p3 in third:
third[p3] += freq
else:
third[p3] = freq
return PositionalBaseUsage(self.BaseUsageClass(first), \
self.BaseUsageClass(second), self.BaseUsageClass(third), self.Info)
def positionalGC(self, purge_unwanted=True):
"""Returns GC, P1, P2 P3. Use purge_unwanted=False to get raw counts."""
p = self.positionalBases(purge_unwanted)
p.normalize()
result = [i['G'] + i['C'] for i in p]
average = sum(result, 0)/3
return [average] + result
def fingerprint(self, which_blocks='quartets', include_mean=True,\
normalize=True):
"""Returns fingerprint data for fingerprint plots.
which_blocks: whether to include only the usual 4-codon quartets (the
default), the split blocks only, or all blocks.
include_mean: whether to include the mean (True).
normalize: whether to normalize so that the quartets sum to 1 (True)
"""
if which_blocks == 'split':
blocks = self.SplitBlocks
elif which_blocks == 'quartets':
blocks = self.SingleAABlocks
elif which_blocks == 'all':
blocks = self.Blocks
else:
raise "Got invalid option %s for which_blocks:\n" % which_blocks+\
" (valid options: 'split', 'quartets', 'all')."
result = []
for b in blocks: #iterates over doublet string
U, C, A, G = [self[b+i] for i in 'UCAG']
all = U+C+A+G
if G+C:
g_ratio = G/(G+C)
else:
g_ratio = 0.5
if A+U:
a_ratio = A/(A+U)
else:
a_ratio=0.5
result.append([g_ratio, a_ratio, all])
result = array(result)
if normalize: #make the shown bubbles sum to 1
sum_ = sum(result[:,-1])
if sum_:
result[:,-1] /= sum_
if include_mean: #calculate mean from all codons
third = self.positionalBases().Third
U, C, A, G = [third[i] for i in 'UCAG']
if G+C:
g_ratio = G/(G+C)
else:
g_ratio = 0.5
if A+U:
a_ratio = A/(A+U)
else:
a_ratio=0.5
result = concatenate((result, array([[g_ratio,a_ratio,1]])))
return result
def pr2bias(self, block):
"""Calculates PR2 bias for a specified block, e.g. 'AC' or 'UU'.
Returns tuple of:
(G/G+C, A/A+T, G/G+A, C/C+T, G/G+T, C/C+A)
If any pair sums to zero, will raise ZeroDivisionError.
block: codon block, e.g. 'AC', 'UU', etc. Any of the 16 doublets.
"""
U, C, A, G = [self[block+i] for i in 'UCAG']
return G/(G+C), A/(A+U), G/(G+A), C/(C+U), G/(G+U), C/(C+A)
def aminoAcids(self, genetic_code=None):
"""Calculates amino acid usage, optionally using a specified code."""
if genetic_code is None:
curr_code = self.GeneticCode
elif isinstance(genetic_code, GenCodeClass):
curr_code = genetic_code
else:
curr_code = GeneticCodes[genetic_code]
aa = {}
for codon, freq in self.items():
curr_aa = curr_code[codon]
if curr_aa in aa:
aa[curr_aa] += freq
else:
aa[curr_aa] = freq
return AminoAcidUsage(aa, self.Info)
class CodonUsage(CodonUsageI, InfoFreqs):
"""Stores frequencies of the 64 codons, mapped to RNA.
This class is convenient but inefficient, since it automatically maps any
lookups to the uppercase RNA alphabet internally. Use UnsafeBaseUsage for
speed when necessary.
"""
Mask = FunctionWrapper(key_to_rna)
RequiredKeys = RnaCodons
BaseUsageClass = BaseUsage
def __init__(self, data=None, Info=None, GeneticCode=None, \
Mask=None, ValueMask=None, Constraint=None):
"""Initializes new CodonUsage with Info and frequency data.
Note: Mask, ValueMask and Constraint are ignored, but must be present
to support copy() because of the ConstrainedContainer interface.
"""
#check if we have a sequence: if so, take it 3 bases at a time
#this will (properly) fail on lists of tuples or anything else where
#the items don't behave like strings.
try:
codons = [''.join(data[i:i+3]) for i in xrange(0, len(data), 3)]
except:
codons = data
super(CodonUsage, self).__init__(codons, Info)
if GeneticCode:
if isinstance(GeneticCode, GenCodeClass):
curr_code = GeneticCode
else:
curr_code = GeneticCodes[GeneticCode]
else:
curr_code = self._default_code
self.__dict__['GeneticCode'] = curr_code
def __getitem__(self, key):
"""Normalizes key and treats T=U."""
key = self.Mask(key)
if len(key) == 2: #pair of bases, e.g. GC for GC content
dup = BaseUsage(self)
dup.normalize()
return sum([dup.get(i,0) for i in key], 0)
else:
return super(CodonUsage, self).__getitem__(key)
class UnsafeCodonUsage(CodonUsageI, UnsafeFreqs):
"""Stores frequencies of the four bases. Must access as RNA.
This class avoids overriding most built-ins, so is much faster than
CodonFreqs (although it is often less convenient).
"""
RequiredKeys = RnaCodons
Info = {} # for interface compatibility with InfoFreqs-based class
Gene=None #for CUTG compatibility
Species=None # for CUTG compaitibility
BaseUsageClass = UnsafeBaseUsage
def __init__(self, data=None, Info=None, GeneticCode=None, \
Mask=None, ValueMask=None, Constraint=None):
"""Initializes new CodonUsage with Info and frequency data.
Note: Mask, ValueMask and Constraint are ignored, but must be present
to support copy() because of the ConstrainedContainer interface.
"""
#check if we have a sequence: if so, take it 3 bases at a time
#this will (properly) fail on lists of tuples or anything else where
#the items don't behave like strings.
try:
codons = [''.join(data[i:i+3]) for i in xrange(0, len(data), 3)]
except:
codons = data or {}
UnsafeFreqs.__init__(self, codons)
#set required keys
for k in self.RequiredKeys:
if k not in self:
self[k] = 0.0
#flatten Info onto self directly for lookups
if Info:
self.__dict__.update(Info)
self.Info = Info or {}
if GeneticCode:
if isinstance(GeneticCode, GenCodeClass):
curr_code = GeneticCode
else:
curr_code = GeneticCodes[GeneticCode]
else:
curr_code = self._default_code
self.GeneticCode = curr_code
class PositionalBaseUsage(Delegator):
"""Stores a BaseUsage for each of the three codon positions."""
def __init__(self, First=None, Second=None, Third=None, Info=None):
"""Returns new PositionalBaseUsage with values for the 3 positions."""
Delegator.__init__(self, Info)
self.__dict__['First'] = First or BaseUsage()
self.__dict__['Second'] = Second or BaseUsage()
self.__dict__['Third'] = Third or BaseUsage()
def _get_info(self):
"""Accessor for Info."""
return self._handler
def _set_info(self, obj):
"""Mutator for Info."""
self._handler = obj
Info = property(_get_info, _set_info)
def __getitem__(self, index):
"""Supports lookups by index."""
if index == 0 or index == -3:
return self.First
elif index == 1 or index == -2:
return self.Second
elif index == 2 or index == -1:
return self.Third
else:
raise IndexError, "PositionalBaseUsage only has 3 positions."
def normalize(self):
"""Normalizes each of the component base usages."""
self.First.normalize()
self.Second.normalize()
self.Third.normalize()
def bases(self):
"""Returns distribution of the four bases, summed over positions."""
sum = BaseUsage(Info=self.Info)
for i in self:
sum +=i
return sum
def codons(self):
"""Returns codon distribution, calculated from positional freqs."""
result = {}
first_copy, second_copy, third_copy = map(Freqs, self)
first_copy.normalize()
second_copy.normalize()
third_copy.normalize()
for c in Codons:
result[c] = first_copy[c[0]] * second_copy[c[1]] * third_copy[c[2]]
return CodonUsage(result, self.Info)
def positionalBases(self):
"""Supports bases/codons/positionalBases/aminoAcids interface."""
return self
def aminoAcids(self, genetic_code=None):
"""Returns amino acid distribution."""
return self.codons().aminoAcids(genetic_code)
class AminoAcidUsage(InfoFreqs):
"""Stores counts ofthe 20 canonical amino acids."""
Mask = FunctionWrapper(upper)
RequiredKeys = dict.fromkeys(AminoAcids)
def bases(self, genetic_code=SGC, codon_usage=_equal_codons):
"""Predicts most likely set of base frequencies.
Optionally uses a genetic code (default: standard genetic code) and
codon usage (default: unbiased codon usage).
"""
result = self.codons(genetic_code, codon_usage).bases()
result.normalize()
return result
def codons(self, genetic_code=SGC, codon_usage=_equal_codons):
"""Predicts most likely set of codon frequencies.
Optionally uses genetic_code (to figure out which codons belong
with each amino acid), and codon_usage (to get most likely codons for
each amino acid). Defaults are the standard genetic code and unbiased
codon frequencies.
"""
result = {}
normalized = Freqs(self)
normalized.normalize()
for aa, aa_freq in normalized.items():
curr_codons = [c.upper().replace('T','U') for c in genetic_code[aa]]
if not curr_codons:
continue #code might be missing some amino acids?
curr_codon_freqs = Numbers([codon_usage[c] for c in curr_codons])
curr_codon_freqs.normalize()
for codon, c_freq in zip(curr_codons, curr_codon_freqs):
result[codon] = c_freq * aa_freq
return CodonUsage(result, self.info, genetic_code)
def positionalBases(self, genetic_code=SGC, codon_usage=_equal_codons):
"""Predicts most likely set of positional base frequencies.
Optionally uses a genetic code (default: standard genetic code) and
codon usage (default: unbiased codon usage).
"""
return self.codons(genetic_code, codon_usage).positionalBases()
def aminoAcids(self):
"""Supports bases/positionalBases/aminoAcids/codons interface."""
return self
class DinucI(object):
"""Provides shared interface for DinucUsage classes.
DinucUsage stores counts of the 16 doublets.
"""
def distance(self, other):
"""Calculates distance between two DinucUsage objects."""
result = 0
for k in self.RequiredKeys:
result += (self[k]-other[k])**2
return sqrt(result)
class DinucUsage(DinucI, InfoFreqs):
"""Stores frequencies of the 16 dinucleotides, mapped to RNA.
This class is convenient but inefficient, since it automatically maps any
lookups to the uppercase RNA alphabet internally. Use UnsafeBaseUsage for
speed when necessary.
"""
Mask = FunctionWrapper(key_to_rna)
RequiredKeys = RnaDinucs
def __init__(self, data=None, Info=None, Overlapping=True, \
GeneticCode=None, Mask=None, ValueMask=None, Constraint=None):
"""Initializes new CodonUsage with Info and frequency data.
Note: Mask, ValueMask and Constraint are ignored, but must be present
to support copy() because of the ConstrainedContainer interface.
"""
#check if we have a sequence: if so, take it 3 bases at a time
#this will (properly) fail on lists of tuples or anything else where
#the items don't behave like strings.
if Mask is not None:
self.Mask = Mask
if ValueMask is not None:
self.ValueMask = ValueMask
try:
data = self.Mask(data)
if Overlapping == '3-1':
range_ = range(2, len(data)-1, 3)
elif Overlapping:
range_ = range(0, len(data)-1)
else:
range_ = range(0, len(data)-1, 2)
dinucs = [''.join(data[i:i+2]) for i in range_]
except:
dinucs = data
super(DinucUsage, self).__init__(dinucs, Info)
def __getitem__(self, key):
"""Normalizes key and treats T=U."""
key = self.Mask(key)
return super(DinucUsage, self).__getitem__(key)
#some useful constants...
EqualBases = BaseUsage()
EqualBases = BaseUsage(_equal_bases)
EqualPositionalBases = PositionalBaseUsage(BaseUsage(_equal_bases),
BaseUsage(_equal_bases), BaseUsage(_equal_bases))
EqualAminoAcids = AminoAcidUsage(_equal_amino_acids) #excl. Stop
EqualCodons = CodonUsage(_equal_codons)
