def __formatList(aminoAcids):
"""Formats 'aminoAcids' into a list of AminoAcid objects."""
# No amino acids result in an empty list
if aminoAcids is None:
return []
# A single AminoAcid is copied and put within a list
elif isinstance(aminoAcids, AminoAcid):
return [AminoAcid(aminoAcids)] # Copy constructor
# A string is converted to a list, based on its
elif isinstance(aminoAcids, str):
if aminoAcids.isupper(): # Multiple Amino Acids in short name mode
return [AminoAcid(aa) for aa in aminoAcids]
else: # A single Amino Acid with any name mode
return [AminoAcid(aminoAcids)]
# A list is copied with all of its items converted to AminoAcid objects
elif isinstance(aminoAcids, list):
return [AminoAcid(aa) for aa in aminoAcids]
# No other supported types
else:
raise TypeError(
"aminoAcids must be a Sequence, list, AminoAcid object, string or None"
)
def getFrequencies(groupValues, groupSizes):
"""
Evaluates the frequencies of AminoAcids within columns of groups in 'groupValues'.
Frequencies are weighted according to group sizes in 'groupSizes'.
Returns two dictionaries and a number:
-'freqPairs' maps pairs of AminoAcids to their frequencies
-'freqSingle' maps single AminoAcids to their frequencies
-'freqSum' is the sum of all frequencies
"""
seqSize = len(groupValues[0]) # Size of the Sequences
groupCount = len(groupSizes) # Number of groups
freqPairs = {} # frequencies of amino acid pairs (fAB)
# Frequencies of single amino acids (fA)
freqSingle = {AminoAcid(aa): 0 for aa in AminoAcid.getAllNames()}
freqSum = 0 # Sum of frequencies sum(fAB)
for col in range(seqSize): # Each column
for groupAIndex in range(groupCount - 1): # Each groupA
groupA = groupValues[groupAIndex]
groupASize = groupSizes[groupAIndex]
for groupBIndex in range(groupAIndex + 1,
groupCount): # Each further groupB
groupB = groupValues[groupBIndex]
groupBSize = groupSizes[groupBIndex]
for aaA, aaACount in groupA[col].items(
): # Each AA from groupA
aaAFreq = aaACount / groupASize # Its frequency within groupA
for aaB, aaBCount in groupB[col].items(
): # Each AA from groupB
aaBFreq = aaBCount / groupBSize # Its frequency within groupB
aaPairFreq = aaAFreq * aaBFreq # Pair frequency
freqSum += aaPairFreq # Sum of all frequencies
freqSingle[aaA] += aaPairFreq / 2
freqSingle[aaB] += aaPairFreq / 2
# Index is unique to this pair
pairIndex = (aaA, aaB) if aaA > aaB else (aaB, aaA)
try:
freqPairs[pairIndex] += aaPairFreq
except:
freqPairs[pairIndex] = aaPairFreq
return freqPairs, freqSingle, freqSum
def __init__(self, path="", description="", ignore=None):
"""
Creates a Score object.
If 'path' is provided, loads the Score values from an iij file.
Otherwise, creates a Score for all possible AminoAcids with values 0.
"""
self._description = description
self._ignore = Sequence(ignore)
self._matrix = []
self._aaOrder = {}
self._aaSequence = Sequence()
# If path is provided, load directly from iij file
if path != "":
with open(path, 'r') as file:
foundAAOrder = False # Have we found the line with the amino acid values and order yet?
for line in file:
if line[0] != "#": # Comments
if not foundAAOrder: # Read aa values and order
for aa in line.split():
self._aaSequence.extend(aa)
self._aaOrder = {
aa: index
for aa, index in zip(
self._aaSequence,
range(len(self._aaSequence)))
}
foundAAOrder = True
else: # Read matrix values
self._matrix.append([int(v) for v in line.split()])
# Otherwise initialize matrix with 0
else:
lineSize = 1
for aa in AminoAcid.getAllNames():
if AminoAcid(aa) not in self._ignore:
self._aaSequence.extend(aa)
self._aaOrder[self._aaSequence[-1]] = lineSize - 1
self._matrix.append([0 for i in range(lineSize)])
lineSize += 1
def __init__(self):
self.structures = "HETC"
self.roc = {(s, r): 0
for s in self.structures for r in ("TP", "TN", "FP", "FN")}
self.minScore, self.maxScore = 0, 0
self.scores = {(classifier, realS): []
for classifier in self.structures
for realS in self.structures}
self.correctPred = 0
self.totalPred = 0
self.neighbourOffset = 8
self.trainings = 0 # Number of trainings (one per AA)
self.strucCount = {s: 0 for s in self.structures}
self.pairCount = {(s, a): 0
for s in self.structures
for a in AminoAcid.getAllNames()}
self.tripletCount = {}
for s in self.structures:
for a in AminoAcid.getAllNames():
for na in AminoAcid.getAllNames(): # Neighbour AA
self.tripletCount[(s, a, na)] = 0
def test_sliceSingleItemIsAminoAcid(self):
prot = Sequence("ABC")
aa = prot[0]
self.assertEquals(aa, AminoAcid("A"))
def test_count(self):
prot = Sequence("ABCAX")
self.assertEquals(prot.count(AminoAcid("A")), 2)
def test_remove(self):
prot = Sequence("TRANKIL")
prot.remove(AminoAcid("A"))
self.assertEquals(prot, Sequence("TRNKIL"))
def __setitem__(self, key, value):
"""Sets value for a slice of the sequence"""
list.__setitem__(self, key, AminoAcid(value))
def canCreateAminoAcid(self, name):
try:
aa = AminoAcid(name)
except ValueError:
self.fail()
def test_hashValuesAreEqual(self):
aa1 = AminoAcid("ala")
aa2 = AminoAcid("A")
self.assertEquals(hash(aa1), hash(aa2))
def test_isTermination(self):
aa = AminoAcid("term")
self.assertTrue(aa.isTermination())
def test_isGap(self):
aa = AminoAcid("gap")
self.assertTrue(aa.isGap())
def test_getName(self):
aa = AminoAcid("ala")
self.assertEquals("A", aa.getName())
def test_getAllNames(self):
for nameMode in ("short", "medium", "long"):
for name in AminoAcid.getAllNames(nameMode):
self.canCreateAminoAcid(name)
def getScore(self, aa1, aa2):
"""
Get the score assigned to AminoAcids 'aa1', 'aa2'.
"""
id1 = self._aaOrder[aa1]
id2 = self._aaOrder[aa2]
if id1 > id2:
return self._matrix[id1][id2]
else:
return self._matrix[id2][id1]
# AA frequencies for complete UniProt database
# from http://web.expasy.org/docs/relnotes/relstat.html, "AMINO ACID COMPOSITION"
uniprob = {
AminoAcid("Ala"): .0826,
AminoAcid("Gln"): .0393,
AminoAcid("Leu"): .0965,
AminoAcid("Ser"): .0660,
AminoAcid("Arg"): .0553,
AminoAcid("Glu"): .0674,
AminoAcid("Lys"): .0582,
AminoAcid("Thr"): .0535,
AminoAcid("Asn"): .0406,
AminoAcid("Gly"): .0708,
AminoAcid("Met"): .0241,
AminoAcid("Trp"): .0109,
AminoAcid("Asp"): .0546,
AminoAcid("His"): .0227,
AminoAcid("Phe"): .0386,
AminoAcid("Tyr"): .0292,