def group_to_reference(fulldict, reference, nonref, structscore, norefseq=False):
nogroup = []
score_structures = ScoreStructures()
for currstruct in nonref:
strscore = structscore
seqscore = 0
bestref = ""
#remove gaps from majority and set as seq1
seq = fulldict[currstruct].majorityConsensus()
seq1 = RnaSequence(''.join(seq).replace('-', ''))
for teststruct in reference:
holdscore = score_structures(currstruct, teststruct)
if holdscore <= strscore:
#remove gaps from majority and set as seq2
seq = fulldict[teststruct].majorityConsensus()
seq2 = RnaSequence(''.join(seq).replace('-', ''))
#compare alignment score. subtract so lower is still better
aln, alnscore = classic_align_pairwise(seq1, seq2, alnscores, -10, -10, False, return_score=True)
if alnscore > seqscore:
strscore = holdscore
seqscore = alnscore
bestref = teststruct
if bestref != "":
#combine the two alignments into one alignment using reference sequence as guide
#refseq must be ungapped to do this without realigning, hence the checks
if norefseq:
#realign all sequences since no refseq available
combinedseqs = fulldict[bestref].degap().addSeqs(fulldict[currstruct].degap())
fulldict[bestref] = align_unaligned_seqs(combinedseqs, RNA, params={"-maxiters": 2, "-diags": True})
continue
#if one refseq is gapless, can easily combine them without realigning
if not fulldict[currstruct].getGappedSeq("refseq").isGapped():
fulldict[bestref].addFromReferenceAln(fulldict[currstruct])
elif not fulldict[bestref].getGappedSeq("refseq").isGapped():
fulldict[currstruct].addFromReferenceAln(fulldict[bestref])
fulldict[bestref] = fulldict[currstruct]
else:
#realign all sequences since both refseqs have gaps
#hacky but it works, need to fix later
fulldict[bestref].Names.remove("refseq")
combinedseqs = fulldict[bestref].degap().addSeqs(fulldict[currstruct].degap())
fulldict[bestref] = align_unaligned_seqs(combinedseqs, RNA)
fulldict[bestref].Names.remove("refseq")
fulldict[bestref].Names.insert(0, "refseq")
fulldict.pop(currstruct)
else:
nogroup.append(currstruct)
score_structures.end()
return fulldict, nogroup
def group_denovo(fulldict, keys, structscore, norefseq=False):
topop = []
score_structures = ScoreStructures()
for pos, currstruct in enumerate(keys):
strscore = structscore
seqscore = 0
bestref = ""
#remove gaps from majority and set as seq1
seq = fulldict[currstruct].majorityConsensus()
seq1 = RnaSequence(''.join(seq).replace('-', ''))
for secpos in range(pos+1, len(keys)):
holdscore = score_structures(currstruct, keys[secpos])
if holdscore <= strscore:
#remove gaps from majority and set as seq2
seq = fulldict[keys[secpos]].majorityConsensus()
seq2 = RnaSequence(''.join(seq).replace('-', ''))
#compare alignment score. Higher is better.
aln, alnscore = classic_align_pairwise(seq1, seq2, alnscores, -10, -10, False, return_score=True)
if alnscore > seqscore:
strscore = holdscore
seqscore = alnscore
bestref = keys[secpos]
if bestref != "":
if norefseq:
#realign all sequences since no refseq available
combinedseqs = fulldict[bestref].degap().addSeqs(fulldict[currstruct].degap())
fulldict[bestref] = align_unaligned_seqs(combinedseqs, RNA)
continue
if not fulldict[currstruct].getGappedSeq("refseq").isGapped():
fulldict[bestref].addFromReferenceAln(fulldict[currstruct])
elif not fulldict[bestref].getGappedSeq("refseq").isGapped():
fulldict[currstruct].addFromReferenceAln(fulldict[bestref])
fulldict[bestref] = fulldict[currstruct]
else:
#realign all sequences since both refseqs have gaps
#hacky but it works, need to fix later
fulldict[bestref].Names.remove("refseq")
combinedseqs = fulldict[bestref].degap().addSeqs(fulldict[currstruct].degap())
fulldict[bestref] = align_unaligned_seqs(combinedseqs, RNA)
fulldict[bestref].Names.remove("refseq")
fulldict[bestref].Names.insert(0, "refseq")
fulldict.pop(currstruct)
topop.append(pos)
topop.sort(reverse=True)
for pos in topop:
keys.pop(pos)
score_structures.end()
return fulldict, keys
def align_order_seqs(seqs, params, outfolder, num, prefix="group_"):
if exists("%s%s%i.fna" % (outfolder, prefix, num)):
return
try:
aln = align_unaligned_seqs(seqs, RNA, params=params)
aln.Names.sort(reverse=True, key=lambda c: count_seqs(c))
with open("%s%s%i.fna" % (outfolder, prefix, num), 'w') as fout:
fout.write(aln.toFasta() + "\n")
except Exception as e:
print("align_order_seqs ERROR: ", format_exc(e))
def bayesfold(seqsin, temperature=37, params=None):
'''Takes in sequences in LoadSeqs readable format and returns
most likely structure from bayesfold.'''
try:
if params == None:
params = {}
aln = align_unaligned_seqs(seqsin, RNA, params=params)
bayesinput = BayesInputWrapper(aln.getSeqNames(),
map(str, aln.iterSeqs()), str(temperature))
bayescalc = BayesCalculation(bayesinput)
bayescalc.run()
struct = str(bayescalc.Alignment.Structures).split()[1]
del bayescalc
del bayesinput
return aln, struct
except Exception, e:
print "BAYESFOLD ERROR: ", e
def bayesfold(seqsin):
'''Runs BayesFold on a set of sequences in MinimalFastaParser format
[(header, seq), (header, seq)] and returns alignment and structure'''
#make sure group has enough sequences before continuing
temperature = 37
seqs = []
aln = align_unaligned_seqs(seqsin, RNA)
for item in aln.Seqs:
seqs.append(str(item))
sequences = RNAAlignment(sequences=seqs)
structures = sequences.fold(temperature, 2, 100)
structalign = str(RNAStructureAlignment(sequences,structures,temperature)).split("\n")
for line in structalign:
if ".." in line:
struct = line
break
return aln, struct
def bayesfold(seqsin, temperature=37, params=None, align=True):
'''Takes in sequences in LoadSeqs readable format and returns
most likely structure from bayesfold.'''
try:
if params is None:
params = {}
if not align:
aln = LoadSeqs(data=seqsin, moltype=RNA, aligned=True)
else:
aln = align_unaligned_seqs(seqsin, RNA, params=params)
bayesinput = BayesInputWrapper(aln.getSeqNames(),
map(str, aln.iterSeqs()),
str(temperature))
bayescalc = BayesCalculation(bayesinput)
bayescalc.run()
struct = str(bayescalc.Alignment.Structures).split()[1]
del bayescalc
del bayesinput
return aln, struct
except Exception, e:
print "BAYESFOLD ERROR: ", format_exc(e)
raise RuntimeError("BAYESFOLD ERROR: ", format_exc(e))
def test_align_unaligned_seqs(self):
"""align_unaligned_seqs should work as expected"""
res = align_unaligned_seqs(self.seqs1, RNA)
self.assertEqual(res.toFasta(), align1)
#load in all haeders for sequences in the clade and match them to their sequence
listin = open(argv[1], 'rU')
fileout = open(folderout + "/" + basenames + "-seqs.fasta", 'w')
listin.readline()
rawseqs = []
tips = []
for line in listin:
header = line.split()[0]
fileout.write(''.join([">", header, "\n", seqs[header], "\n"]))
rawseqs.append((header, seqs[header]))
tips.append(header)
fileout.close()
print "Aligning seqs using muscle with -diags"
seqs = LoadSeqs(data=rawseqs, moltype=RNA, aligned=False)
aln = align_unaligned_seqs(seqs, RNA, {"-diags": True})
fileout = open(folderout + "/" + basenames + "-seqsaligned.fasta", 'w')
fileout.write(str(aln))
fileout.close()
print "Folding sequences"
#get subtree of the clade being folded to pass to PPfold
tr = LoadTree(argv[3])
sub_tree = tr.getSubTree(tips, keep_root=True)
filesubtree = open(folderout + "/" + basenames + "-subtreeDistances.nwk", 'w')
filesubtree.write(sub_tree.getNewick(with_distances=True))
filesubtree.close()
filesubtree = open(folderout + "/" + basenames + "-subtree.nwk", 'w')
filesubtree.write(sub_tree.getNewick(with_distances=False))
#call PPfold with aligned sequences and subtree
args = ["java", "-jar", PPFOLDDIR + "PPfold.jar", folderout + "/" + basenames + "-seqsaligned.fasta", "--outputd", folderout]
else:
print "Clusters previously folded"
clusters.clear()
del clusters
if not exists(otufolder + "clusters_aln.fasta"):
startaln = time()
if not args.nr:
#add refseq and align all sequences for later seq/struct comparisons
refin = open(args.f + "refseq.fasta")
crap, refseq = MinimalFastaParser(refin).next()
refin.close()
for struct in structgroups:
if not args.nr:
structgroups[struct].append(("refseq", refseq))
structgroups[struct] = align_unaligned_seqs(structgroups[struct], RNA)
if not args.nr:
#hacky way to make sure refseq first, will need to fix later
structgroups[struct].Names.remove("refseq")
structgroups[struct].Names.insert(0, "refseq")
#write that shit to a file to save time if rerun needed
cout = open(otufolder + "clusters_aln.fasta", 'w')
for struct in structgroups:
cout.write(">%s\n%s\n" % ("newcluster", struct))
cout.write(structgroups[struct].toFasta() + "\n")
cout.close()
print "Aligned all clusters: " + str((time() - startaln)/60) + " min"
else:
cin = open(otufolder + "clusters_aln.fasta")
currclust = []
currstruct = ""
def test_align_unaligned_seqs(self):
"""align_unaligned_seqs should work as expected"""
res = align_unaligned_seqs(self.seqs1, RNA)
self.assertEqual(res.toFasta(), align1)
print c[0] + "\t" + str(c[1])
print str(len(clusters)) + " clusters"
#print that shit to file
clustfiles = []
for cnum,cinfo in enumerate(countarray):
if not exists(''.join([argv[2],"cluster", str(cnum), "_", str(cinfo[1]), ".txt"])):
clusters[cinfo[0]].sort(reverse=True,key=lambda count: int(count[0].split('_')[1]))
cout = open(''.join([argv[2],"cluster", str(cnum), "_", str(cinfo[1]), ".txt"]), 'w')
for seq in clusters[cinfo[0]]:
cout.write(">%s\n%s\n" % seq)
cout.close()
clustfiles.append(''.join([argv[2],"cluster", str(cnum), "_", str(cinfo[1]), ".txt"]))
for cfile in clustfiles:
print cfile
seqs = LoadSeqs(cfile, moltype=RNA, format='fasta')
aln = align_unaligned_seqs(seqs,RNA)
cout = open(cfile[:cfile.rfind(".")] + "_aln.txt", 'w')
cout.write(aln.toFasta())
cout.close()
fin = open(cfile[:cfile.rfind(".")] + "_aln.txt")
seqs = read_seq_data(fin)
fin.close()
data = LogoData.from_seqs(seqs)
options = LogoOptions()
options.title = "cluster logo"
format = LogoFormat(data, options)
fout = open(cfile[:cfile.rfind(".")] + ".eps", 'w')
eps_formatter(data, format, fout)
fout.close()
def phyl_tree():
mytree = nj.nj(d.getPairwiseDistances())
print ("\n\n")
print mytree.asciiArt()
al = LoadSeqs("cytc.fasta", moltype=PROTEIN, interleaved=False)
d = distance.EstimateDistances(al, submodel = JTT92())
d.run()
sys.stdout = open("cytc distances.txt", "w")
print d
phyl_tree()
al = LoadSeqs("mtdna.fasta", moltype=DNA, interleaved=True, aligned=False)
d = distance.EstimateDistances(al, submodel = JC69())
d.run()
sys.stdout = open("mtdna distances.txt", "w")
print d
phyl_tree()
seqs = LoadSeqs("cytb.fasta", moltype=PROTEIN, aligned=False)
al = align_unaligned_seqs(seqs,PROTEIN)
dcalc = distance.EstimateDistances(al, submodel = JTT92())
dcalc.run(show_progress = True)
d = dcalc.getPairwiseDistances()
tree=nj.nj(d)
sys.stdout = open("cytb distances.txt", "w")
print dcalc
print '\n\n'
print tree.asciiArt()
#phyl_tree()
