def align(seq1, seq2, debug=False):
flat1 = seq.seq1(''.join(seq1)).replace('X', '-')
flat2 = seq.seq1(''.join(seq2)).replace('X', '-')
flats = [flat1, flat2]
# aligning 2 to 1 seems to give better results
align = pairwise2.align.localxs(flat2,
flat1,
-1000,
-1000,
one_alignment_only=True)
start = align[0][3]
offset = [0, 0]
# compute how many gaps had to be inserted at beginning to align
for i in range(2):
assert len(align[0][0]) == len(align[0][1])
for j in range(len(align[0][0])):
# account for the fact that 2 and 1 are switched in alignment results
# if there is a gap in 1
if align[0][(i + 1) % 2][j] == '-':
# but not the other
if flats[i][j - offset[i]] != '-':
offset[i] += 1
else:
break
if debug:
print(
pairwise2.format_alignment(flat2[offset[0]:], flat1[offset[1]:],
10, 0,
len(flat1) - offset[1]))
return -offset[0], -offset[1]
def var3_to_var1(var3):
refAA3 = variant.get_refAA(var3)
newAA3 = variant.get_newAA(var3)
pos = variant.get_pos(var3)
refAA1 = "*" if refAA3 == "*" else SeqUtils.seq1(refAA3)
# refAA1 = SeqUtils.seq1(refAA3)
newAA1 = "*" if newAA3 == "*" else SeqUtils.seq1(newAA3)
# newAA1 = SeqUtils.seq1(newAA3)
var1 = ''.join([refAA1, str(pos), newAA1])
return var1
def get_distances(res_pairs, get_coords):
''' Get distances for all pairs of residues between two chains
res_pairs: generator over tuples ((res_a, res_b), ...)
get_coords: function to get residue coordinates
Returns a list over 5-tuples: [(resn_a, resn_b, aa_a, aa_b, dist), ...]
'''
return [(res_a.id[1], res_b.id[1],
distances.calc_residue_distance(res_a, res_b, get_coords),
SeqUtils.seq1(res_a.resname), SeqUtils.seq1(res_b.resname))
for (res_a, res_b) in res_pairs]
def get_sequences(pdb_id, chain=None):
'''Gets the sequences in a PDB file.'''
return [SeqUtils.seq1(''.join([residue.get_resname()
for residue in chn
if 'CA' in residue.child_dict]))
for chn in get_structure(pdb_id).get_chains()
if chain is None or chain == chn.get_id()]
def extractPDBdata(structure, adjustChains, substitutionData, verbose):
print('Extracting atoms details from PDB...')
pdbData = {}
for model in structure:
for chain in model:
chainID = chain.get_id()
if chainID in adjustChains:
pdbData[chainID] = {}
residueID = 0
for residue in chain:
residueName = SeqUtils.seq1(residue.get_resname())
if residueName != substitutionData[chainID][residueID][0]:
continue
(heteroFlag, sequenceID, insertionCode) = residue.get_id()
if heteroFlag != ' ':
continue
value = substitutionData[chainID][residueID][1]
if value != "-":
pdbData[chainID][sequenceID] = value
if verbose:
print("Chain: " + chainID + "\t residue: " + residueName + " " + str(sequenceID) + "\t value: " + value)
residueID += 1
if (residueID >= len(substitutionData[chainID])):
break
print('OK')
return pdbData
def get_bfactors(self, chain_id):
'''
Input:
self: Use Biopython.PDB structure which has been stored in an object variable
chain_id : String (usually in ['A','B', 'C' ...]. The number of chains
depends on the specific protein and the resulting structure)
Return:
Return the B-Factors for all residues in a chain of a Biopython.PDB structure.
The B-Factors describe the mobility of an atom or a residue.
In a Biopython.PDB structure B-Factors are given for each atom in a residue.
Calculate the mean B-Factor for a residue by averaging over the B-Factor
of all atoms in a residue.
Sometimes B-Factors are not available for a certain residue;
(e.g. the residue was not resolved); insert np.nan for those cases.
Finally normalize your B-Factors using Standard scores (zero mean, unit variance).
You have to use np.nanmean, np.nanvar etc. if you have nan values in your array.
The returned data structure has to be a numpy array rounded again to integer.
'''
bf = [
np.mean([a.get_bfactor() for a in r])
for r in self.structure[chain_id]
if SeqUtils.seq1(r.get_resname()) != 'X'
]
meanBf = np.mean(bf)
stdBf = np.std(bf)
bf = [(b - meanBf) / stdBf for b in bf]
return np.array(bf).astype(np.int) # return rounded (integer) values
def get_distances(res_pairs, get_coords):
''' Get distances for all pairs of residues between two chains
res_pairs: generator over tuples ((res_a, res_b), ...)
get_coords: function to get residue coordinates
Returns a list over 5-tuples: [(resn_a, resn_b, aa_a, aa_b, dist), ...]
'''
return [
(res_a.id[1], res_b.id[1],
distances.calc_residue_distance(res_a, res_b, get_coords),
SeqUtils.seq1(res_a.resname), SeqUtils.seq1(res_b.resname)
)
for (res_a, res_b)
in res_pairs
]
def generate_wt_seqs(peptides):
wt_dict = {}
r = re.compile("([a-zA-Z]+)([0-9]+)([a-zA-Z]+)")
d_pattern = re.compile("([a-zA-Z]+)([0-9]+)")
for x in peptides:
trans = x.get_all_transcripts()
for t in trans:
mut_seq = [a for a in x]
protein_pos = x.get_protein_positions(t.transcript_id)
not_available = False
variant_available = False
for p in protein_pos:
variant_dic = x.get_variants_by_protein_position(
t.transcript_id, p)
variant_available = bool(variant_dic)
for key in variant_dic:
var_list = variant_dic[key]
for v in var_list:
mut_syntax = v.coding[t.transcript_id.split(':')
[0]].aaMutationSyntax
if v.type in [3, 4, 5] or '?' in mut_syntax:
not_available = True
elif v.type in [1]:
m = d_pattern.match(mut_syntax.split('.')[1])
wt = SeqUtils.seq1(m.groups()[0])
mut_seq.insert(key, wt)
elif v.type in [2]:
not_available = True
else:
m = r.match(mut_syntax.split('.')[1])
if m is None:
not_available = True
else:
wt = SeqUtils.seq1(m.groups()[0])
mut_seq[key] = wt
if not_available:
wt_dict['{}_{}'.format(str(x), t.transcript_id)] = np.nan
elif variant_available:
wt_dict['{}_{}'.format(str(x),
t.transcript_id)] = ''.join(mut_seq)
return wt_dict
def changeBfactors(structure, adjustChains, substitutionData, verbose,
noDataValue):
print('Changing b-factors of atoms in PDB...')
for model in structure:
for chain in model:
chainID = chain.get_id()
if chainID in adjustChains:
residueID = 0
for residue in chain:
residueName = SeqUtils.seq1(residue.get_resname())
if residueName != substitutionData[chainID][residueID][0]:
if verbose:
print("WARNING: Unexpected residue " +
str(residueID) + " in chain " + str(chainID))
print("Expected residue " +
substitutionData[chainID][residueID][0] +
" got " + residueName)
continue
(heteroFlag, sequenceID, insertionCode) = residue.get_id()
if heteroFlag != ' ':
continue
for atom in residue:
value = float(substitutionData[chainID][residueID][1])
if atom.is_disordered():
atom.__class__ = Atom.DisorderedAtom
for altloc in atom.disordered_get_id_list():
a = atom.disordered_get(altloc)
if verbose:
print("Chain: " + chainID +
"\t residue: " + residueName +
"\t atom: " + str(a.get_full_id()) +
" \t b-factor: " +
str(a.get_bfactor()) + " => " +
str(value))
a.set_bfactor(value)
else:
if verbose:
print("Chain: " + chainID + "\t residue: " +
residueName + "\t atom: " +
str(atom.get_full_id()) +
" \t b-factor: " +
str(atom.get_bfactor()) + " => " +
str(value))
atom.set_bfactor(value)
residueID += 1
if (residueID >= len(substitutionData[chainID])):
break
else:
for residue in chain:
for atom in residue:
atom.set_bfactor(float(noDataValue))
print('OK')
return structure
def get_number_of_water_molecules(self, chain_id):
'''
Input:
self: Use Biopython.PDB structure which has been stored in an object variable
chain_id : String (usually in ['A','B', 'C' ...]. The number of chains
depends on the specific protein and the resulting structure)
Return:
Return the number of water molecules of a given chain (chain_id)
in a Biopython.PDB structure as an integer.
'''
n_waters = 0
for r in self.structure[chain_id]:
n_waters += SeqUtils.seq1(r.get_resname()) == "X"
return n_waters
def get_dihedral( residue_list ):
'''
returns phi and psi angles of a residue and the amino acid sidechain present
residue_list - []Bio.PDB.Residue - list of 3 *hopefully* continuous residues
'''
for one, two in zip( residue_list[:-1], residue_list[1:] ):
if ( two.get_id()[1] - one.get_id()[1] ) != 1:
raise BackboneError( "Discontinuous residues", two.get_id()[1] )
atoms = (
{"C": False},
{"N": False,
"CA": False,
"C": False},
{"N": False}
)
for i, residue in enumerate( residue_list ):
if i == 1:
res_name = SeqUtils.seq1( residue.get_resname() )
if not is_aa( res_name ):
raise BackboneError( "Not a valid amino acid", residue.get_id()[1] )
for atom in residue.get_unpacked_list():
if atom.name in atoms[i].keys():
atoms[i][ atom.name ] = atom.get_vector()
if False in map( check_dict, atoms ):
raise BackboneError( "Missing backbone atoms", residue.get_id()[1] )
dihedrals = [
PDB.calc_dihedral( atoms[0]["C"], atoms[1]["N"], atoms[1]["CA"], atoms[1]["C"] ), #phi
PDB.calc_dihedral( atoms[1]["N"], atoms[1]["CA"], atoms[1]["C"], atoms[2]["N"] ) #psi
]
return ( dihedrals, res_name )
def get_sequence(self, chain_id):
'''
Input:
self: Use Biopython.PDB structure which has been stored in an object variable
chain_id : String (usually in ['A','B', 'C' ...]. The number of chains
depends on the specific protein and the resulting structure)
Return:
Return the amino acid sequence (single-letter alphabet!) of a given chain (chain_id)
in a Biopython.PDB structure as a string.
'''
sequence = ''
for r in self.structure[chain_id]:
s = SeqUtils.seq1(r.get_resname())
if s != "X":
sequence = sequence + s
return sequence
def get_codon_usage(cuspp):
"""
Creates codon usage table.
:param cuspp: path to cusp generated file
:returns: codon usage table (pandas dataframe)
"""
# get codon usage stats
dcodonusage = pd.read_csv(cuspp, sep='\t', header=5)
cols = ''.join(dcodonusage.columns.tolist()).split(' ')
dcodonusage.columns = [cols[-1]]
dcodonusage.index.names = cols[:-1]
dcodonusage = dcodonusage.reset_index().set_index('Codon')
dcodonusage['amino acid'] = [SeqUtils.seq1(s) for s in dcodonusage['#AA']]
return dcodonusage
def get_sequence(self, chain_id):
"""
Input:
self: Use Biopython.PDB structure which has been stored in an object variable
chain_id : String (usually in ['A','B', 'C' ...]. The number of chains
depends on the specific protein and the resulting structure)
Return:
Return the amino acid sequence (single-letter alphabet!) of a given chain (chain_id)
in a Biopython.PDB structure as a string.
"""
sequences = list(self.structure.get_chains())
for i in sequences:
if i.id == chain_id:
ret = ""
for j in i.get_list():
ret += SeqUtils.seq1(j.resname)
return ret.replace("X", "")
return None
def Chain_to_SeqRecord(chain):
''' Generates a SeqRecord from a Chain entity.
chain: a Bio.PDB.Chain object
Keeps only residues with blank flags (eg. no HET residues).
Returns seqr: a Bio.SeqRecord object with a list of resnums saved in
its letter_annotations['resnum'].
'''
aas = ''
resns = list()
for res in get_nonhet_residues(chain):
aas += SeqUtils.seq1(res.get_resname()) # get 1-letter resname
resns += [res.id[1]]
seqr = SeqRecord.SeqRecord(Seq.Seq(aas), id = chain.id,
letter_annotations = {"resnum": resns})
return seqr
def Chain_to_SeqRecord(chain):
''' Generates a SeqRecord from a Chain entity.
chain: a Bio.PDB.Chain object
Keeps only residues with blank flags (eg. no HET residues).
Returns seqr: a Bio.SeqRecord object with a list of resnums saved in
its letter_annotations['resnum'].
'''
aas = ''
resns = list()
for res in get_nonhet_residues(chain):
aas += SeqUtils.seq1(res.get_resname()) # get 1-letter resname
resns += [res.id[1]]
seqr = SeqRecord.SeqRecord(Seq.Seq(aas),
id=chain.id,
letter_annotations={"resnum": resns})
return seqr
def parse_change(self, change):
dref, dalt = change.split("->")[0].strip().split("/")
dref = SeqUtils.seq1(dref)
dalt = SeqUtils.seq1(dalt)
return (dref, dalt)
if verbose: print '\n', linia
asequence = sequence_lines[i][:f_length].replace('.', 'n')
bsequence = sequence[start - 1:end]
if strand == "-":
bsequence = str(Seq(bsequence).reverse_complement())
mfe = RNA.energy_of_structure(bsequence, f_struct, 0)
if verbose:
print f_struct, mfe
print asequence
print bsequence
score = str(-1 * mfe)
# score = elementy_linii[3] #this was actually wrong - it is relative anticodon position!
gname = elementy_linii[1].split("-")[1]
label = SeqUtils.seq1(gname)
anticodon_position = int(elementy_linii[3]) - 1
# anticodon = elementy_linii[4]
if True: ###gname in ['Ser','Leu','Met']:
# print elementy_linii[4]
if gname in ['Ser', 'Arg', 'Gly'
] and elementy_linii[4][2:4] == 'ct':
label = label + '2'
if ttable_id == '4' and gname == 'Arg' and elementy_linii[4][
2:4] == 'cg':
label = label + '1'
if ttable_id == '13' and gname == 'Gly' and elementy_linii[4][
2:4] == 'cc':
label = label + '1'
if gname == 'Ser' and elementy_linii[4][2:4] == 'ga':
label = label + '1'
def load(self):
'''
GeneID Drug GeneName NucleotidePosition Polymorphism EstimatedCodonPosition ReportedCodonPosition AminoAcid
0 Rv3795 EMB embB 1121-1122 GGC/GTG 374 374 Gly/Val
1 Rv3795 EMB embB 1123-1124 CCG/GCG 375 375 Pro/Ala
'''
cols = ['ID', 'GeneID', 'SeqNo', 'Drug', 'GeneName', 'ApprovalLevel', 'AddedByUser', 'PrimaryReference',
'ReviewReferece',
'SecondaryReference', 'NucleotidePosition', 'Polymorphism', 'EstimatedCodonPosition',
'ReportedCodonPosition',
'AminoAcid', 'Note', 'TimePeriod', 'StudyPopulation', 'Country', 'MolecularDetectionMethod',
'GeneCoverage',
'ResistancePattern', 'MIC', 'SusceptibilityTestingMethod', 'RTotalIsolates', 'RSIsolatesWMutation',
'AdditionalMutations',
'HighQuality', 'PMID', 'OverallSequentialNumber', 'NoHC', 'Temp1', 'Temp2', 'Temp3', 'Temp4', 'Temp5',
'MutationType']
self._df = pd.DataFrame()
errors = []
i = 0
rvs = []
with open(self.csv_db_path, "r") as handle:
lines = handle.readlines()
_log.debug("Total lines count: " + str(len(lines)))
for idx, line in enumerate(lines):
i = i + 1
if i < 3:
continue
split = line.split(",")
if len(split) >= 14:
dict_snp = {"line": line}
j = 0
for col in cols:
dict_snp[col] = split[j]
j = j + 1
if (not (("/" in dict_snp["Polymorphism"]) or ("ins" in dict_snp["Polymorphism"]) or (
"del" in dict_snp["Polymorphism"]))
and
not (("/" in dict_snp["AminoAcid"]) or ("ins" in dict_snp["AminoAcid"]) or (
"del" in dict_snp["AminoAcid"]))
):
print dict_snp["ID"]
self._df = self._df.append(dict_snp, ignore_index=True)
rvs.append(dict_snp["GeneID"])
# i = i -1
# if not i:
# break
# (rv,nucleotide,polimorphism,codon_aa,change_aa,drug) = (split[1],split[10],split[11],split[14],split[12],split[3])
# if isinstance( codon_aa,int):
# snps.append((rv,nucleotide,polimorphism,codon_aa,change_aa,drug))
# else:
# errors.append(line)
else:
errors.append(line)
self._df["codon"] = map(self._codon_position, [x for i, x in self._df.iterrows()])
self._df["change"] = map(self._mutated_aa, [x for i, x in self._df.iterrows()])
self._df["ref"] = map(lambda x: bpsutils.seq1(x[0]) if x else None, self._df["change"])
self._df["mutation"] = map(lambda x: bpsutils.seq1(x[1]) if x else None, self._df["change"])
self._df["mut_type"] = map(self._mutation_type, [x for i, x in self._df.iterrows()])
self._df["rv"] = map(lambda x: x.lower(), self._df["GeneID"])
self._df["gene"] = map(lambda x: x.lower(), self._df["GeneName"])
self._df["nu_pos"] = map(self.nu_pos, self._df["NucleotidePosition"])
self._df["nu_ref"] = map(lambda x: x.split("/")[0].strip(), self._df["Polymorphism"])
self._df["nu_alt"] = map(self.nu_alt, self._df["Polymorphism"])
self._df["raw"] = line
self._df["rna"] = map(lambda x: True if x == None else False, self._df["change"])
_log.info("SNPs loaded:" + str(len(self._df)))
_log.info("Errors: " + str(len(errors)))
_log.info("RV count: " + str(len(set(rvs))))
def load_in_sndg(self, organism="H37Rv"):
from SNDG.BioMongo.Model.Protein import Protein
from SNDG.BioMongo.Model.Feature import Feature, Location
from SNDG.BioMongo.Model.SeqCollection import SeqCollection
from SNDG.BioMongo.Model.SeqColDruggabilityParam import SeqColDruggabilityParamTypes, SeqColDruggabilityParam
from bson.objectid import ObjectId
search_params = [("resistance", "Associated with resistance", "variant-db",
SeqColDruggabilityParamTypes.value, ["true", "false"], "true", "equal", "avg")
]
search_params = search_params + [
(x, "Associated with " + x + " resistance", "variant-db",
SeqColDruggabilityParamTypes.value, ["true", "false"], "true", "equal", "avg")
for x in TBDream.drugs
]
Protein.objects(organism=organism).update(__raw__={"$pull": {"features": {"type": "tbdream"}}})
collection = SeqCollection.objects(name=organism).get()
for name, description, target, _type, options, defaultValue, defaultOperation, defaultGroupOperation in search_params:
Protein.objects(organism=organism).update(__raw__={"$set": {"search." + name: False}})
if not collection.has_druggability_param(name):
dp = SeqColDruggabilityParam(name=name, description=description, target=target,
type=_type, uploader="demo")
dp.options = options
dp.defaultValue = defaultValue
dp.defaultOperation = defaultOperation
dp.defaultGroupOperation = defaultGroupOperation
collection.druggabilityParams.append(dp)
collection.save()
for rv, rows in self._df.groupby("rv"):
prot = list(Protein.objects(organism=organism, gene__iexact=rv))
if prot:
prot = prot[0]
for _, r in rows.iterrows():
mut = None
if r.change:
change = str(r.change[0]) + "/" + str(r.change[1])
mut = SeqUtils.seq1(r.change[1])
else:
change = r.AminoAcid
if math.isnan(r.codon):
try:
pos = int(r.AminoAcid)
except:
_log.warn("couldnt find the variant position")
continue
else:
pos = int(r.codon)
try:
res, t = r.RTotalIsolates.strip().split("/")
r_div_total_coef = int(res) * 1.0 / int(t)
r_div_total = r.RTotalIsolates.strip()
except:
r_div_total = None
r_div_total_coef = None
quals = {
"drug": r.Drug,
"change": change,
"gene": r.GeneID,
"pattern": r.ResistancePattern,
"additional": r.AdditionalMutations,
"r_div_total": r_div_total,
"r_div_total_coef": r_div_total_coef,
"mic": r.MIC}
if mut:
quals["mut"] = mut
fvariant = Feature(_id=ObjectId(), location=Location(start=pos, end=pos), type="tbdream",
identifier="TBDream id " + r.ID,
qualifiers=quals)
prot.features.append(fvariant)
prot.search.resistance = True
prot.search[r.Drug] = True
prot.save()