def weblogoPOIM(logofile, poim, max_len):
"""instead of plotting the POIM heatmap, create a weblogo from the 1st-degree poim"""
warnings.filterwarnings('ignore', ' This call to matplotlib.use()*')
from corebio.data import rna_letters, dna_letters, amino_acid_letters
from weblogolib import LogoData, LogoOptions, LogoFormat, classic, png_print_formatter
#print "WEBLOGO!"
#print "Writing ", logofile
#print poim[0]
positive_logo = []
negative_logo = []
for i in xrange(len(poim[0])):
positive_logo.append([])
negative_logo.append([])
for j in xrange(len(poim[0][i])):
if poim[0][i][j] < 0:
positive_logo[i].append(0)
negative_logo[i].append(poim[0][i][j] * -10000)
else:
negative_logo[i].append(0)
positive_logo[i].append(poim[0][i][j] * 1000)
#print "Positive logo: ", positive_logo
#print "Negative logo: ", negative_logo
pos_data = LogoData.from_counts('ACGT', numpy.array(positive_logo).T, None)
neg_data = LogoData.from_counts("ACGT", numpy.array(negative_logo).T, None)
neg_opt = LogoOptions()
neg_opt.fineprint += " from KIRMES POIM data"
#logoopt.number_interval = 5
neg_opt.small_fontsize = 4
neg_opt.title_fontsize = 8
neg_opt.scale_width = False
title = os.path.split(logofile)[1]
title = title[:title.rfind(".")]
if "_" in title:
title = title[title.rfind("_") + 1:]
neg_opt.logo_title = title + " Negative Logo"
neg_format = LogoFormat(neg_data, neg_opt)
pos_opt = LogoOptions()
#pos_opt.show_ends = True
pos_opt.scale_width = False
pos_opt.logo_title = title + " Positive Sequence Logo"
pos_opt.show_fineprint = False
pos_opt.color_scheme = classic
pos_format = LogoFormat(pos_data, pos_opt)
neg_logo = open(logofile + "n.png", 'w')
png_print_formatter(neg_data, neg_format, neg_logo)
neg_logo.close()
pos_logo = open(logofile + "p.png", 'w')
png_print_formatter(pos_data, pos_format, pos_logo)
pos_logo.close()
concatPNG(logofile, (logofile + "p.png", logofile + "n.png"))
os.remove(logofile + "n.png")
os.remove(logofile + "p.png")
def _build_logodata(options):
if options.input_parser != "transfac":
seqs = read_seq_data(options.fin,
options.input_parser.read,
alphabet=options.alphabet,
ignore_lower_case=options.ignore_lower_case)
if options.reverse:
seqs = SeqList([s.reverse() for s in seqs], seqs.alphabet)
if options.complement:
seqs = SeqList([Seq(s, seqs.alphabet).complement() for s in seqs],
seqs.alphabet)
prior = parse_prior(options.composition, seqs.alphabet, options.weight)
data = LogoData.from_seqs(seqs, prior)
else:
from corebio.matrix import Motif
if options.ignore_lower_case:
raise ValueError(
"error: option --ignore-lower-case incompatible with matrix input"
)
#FIXME : implement
if options.reverse:
raise ValueError(
"error: option --reverse incompatible with matrix input")
#FIXME : implement
if options.complement:
raise ValueError(
"error: option --complement incompatible with matrix input")
motif = Motif.read_transfac(options.fin, alphabet=options.alphabet)
prior = parse_prior(options.composition, motif.alphabet,
options.weight)
data = LogoData.from_counts(motif.alphabet, motif, prior)
return data
def write_weblogo(self, filepath):
matrix_tuple = []
for distribution in self.values:
matrix_tuple.append(tuple(distribution))
dataArray = np.array(tuple(matrix_tuple))
alph = Alphabet(''.join(self.alphabet))
weblogoData = LogoData.from_counts(alph, dataArray)
weblogoOptions = LogoOptions(color_scheme=classic)
weblogoOptions.title = "PWM"
weblogoFormat = LogoFormat(weblogoData, weblogoOptions)
weblogo_file = open(filepath, 'w')
weblogo_file.write(png_print_formatter(weblogoData, weblogoFormat))
weblogo_file.close()
def outputMotif(theta_motif, theta_background_matrix, lambda_motif, logev, k, outstr):
from weblogolib import LogoData, LogoOptions, LogoFormat, png_formatter, eps_formatter, unambiguous_dna_alphabet
_pv_format = "%3.1fe%+04.0f"
f_string = sprint_logx(log(lambda_motif), 1, _pv_format)
g_string = sprint_logx(logev, 1, _pv_format)
print(("Motif {0:s} had a fraction of {1:s}").format(str(k), f_string))
print(("Motif {0:s} had an E-value of {1:s}").format(str(k), g_string))
print 'Saving motif as a png...'
data = LogoData.from_counts(counts=theta_motif,alphabet=unambiguous_dna_alphabet)#,prior=theta_background_matrix[0])#Does prior mess things up?
options = LogoOptions()
options.title = 'Motif'
forma = LogoFormat(data, options)
fout = open(outstr + "Motif_" + str(k) + '.png', 'w')
png_formatter(data, forma, fout)
fout.close()
print 'Saving motif as an eps...'
fout = open(outstr + "Motif_" + str(k) + '.eps', 'w')
eps_formatter(data, forma, fout)
fout.close()
def outputMotif(theta_motif, theta_background_matrix, lambda_motif, logev, k, outstr):
from weblogolib import LogoData, LogoOptions, LogoFormat, png_formatter, eps_formatter, unambiguous_dna_alphabet
_pv_format = "%3.1fe%+04.0f"
f_string = sprint_logx(log(lambda_motif), 1, _pv_format)
g_string = sprint_logx(logev, 1, _pv_format)
print(("Motif {0:s} had a fraction of {1:s}").format(str(k), f_string))
print(("Motif {0:s} had an E-value of {1:s}").format(str(k), g_string))
print 'Saving motif as a png...'
data = LogoData.from_counts(counts=theta_motif,alphabet=unambiguous_dna_alphabet)#,prior=theta_background_matrix[0])#Does prior mess things up?
options = LogoOptions()
options.title = 'Motif'
forma = LogoFormat(data, options)
fout = open(outstr + "Motif_" + str(k) + '.png', 'w')
png_formatter(data, forma, fout)
fout.close()
print 'Saving motif as an eps...'
fout = open(outstr + "Motif_" + str(k) + '.eps', 'w')
eps_formatter(data, forma, fout)
fout.close()
def createSeqLogo(pfm, filename, fformat='eps'):
"""Create sequence logo for an individual cRBM motif.
Parameters
-----------
pfm : numpy-array
2D numpy array representing a PFM. See :meth:`CRBM.getPFMs`
path : str
Output folder.
fformat : str
File format for storing the sequence logos. Default: 'eps'.
"""
alph = Alphabet('ACGT')
weblogoData = LogoData.from_counts(
alph, pfm.T) #, c)#, learner.c.get_value().reshape(-1))
weblogoOptions = LogoOptions(color_scheme=classic)
weblogoFormat = LogoFormat(weblogoData, weblogoOptions)
content = formatters[fformat](weblogoData, weblogoFormat)
f = open(filename, "wb")
f.write(content)
f.close()
motif_flag = True
except ValueError, motif_err :
# Failed reading Motif, try reading as multiple sequence data.
seqs = read_seq_data(fin,
options.input_parser.read,
alphabet=options.alphabet,
ignore_lower_case = options.ignore_lower_case)
if motif_flag :
if options.ignore_lower_case:
raise ValueError("error: option --ignore-lower-case incompatible with matrix input")
if options.reverse: motif.reverse()
if options.complement: motif.complement()
prior = parse_prior( options.composition,motif.alphabet, options.weight)
data = LogoData.from_counts(motif.alphabet, motif, prior)
else :
if options.reverse:
seqs = SeqList([s.reverse() for s in seqs], seqs.alphabet)
if options.complement :
seqs= SeqList( [Seq(s,seqs.alphabet).complement() for s in seqs], seqs.alphabet)
prior = parse_prior( options.composition,seqs.alphabet, options.weight)
data = LogoData.from_seqs(seqs, prior)
return data
seqs = read_seq_data(fin,
options.input_parser.read,
alphabet=options.alphabet,
ignore_lower_case=options.ignore_lower_case)
if motif_flag:
if options.ignore_lower_case:
raise ValueError(
"error: option --ignore-lower-case incompatible with matrix input"
)
if options.reverse: motif.reverse()
if options.complement: motif.complement()
prior = parse_prior(options.composition, motif.alphabet,
options.weight)
data = LogoData.from_counts(motif.alphabet, motif, prior)
else:
if options.reverse:
seqs = SeqList([s.reverse() for s in seqs], seqs.alphabet)
if options.complement:
seqs = SeqList([Seq(s, seqs.alphabet).complement() for s in seqs],
seqs.alphabet)
prior = parse_prior(options.composition, seqs.alphabet, options.weight)
data = LogoData.from_seqs(seqs, prior)
return data
def _build_logoformat(logodata, opts):
options.alphabet = None
seqs = read_seq_data(
fin, options.input_parser.read, alphabet=options.alphabet, ignore_lower_case=options.ignore_lower_case
)
if motif_flag:
if options.ignore_lower_case:
raise ValueError("option --ignore-lower-case incompatible with matrix input")
if options.reverse:
motif.reverse()
if options.complement:
motif.complement()
if not isCodon:
prior, compos = parse_prior(fin_compos, motif.alphabet, fin_weight)
data = LogoData.from_counts(motif.alphabet, motif, options.stats_func, prior, compos, second_data)
else:
raise ValueError("option --sequence-type 'codon' incompatible with matrix input")
else:
if options.codon_frame < 0 and isCodon:
options.reverse = True
options.complement = True
if options.reverse:
seqs = SeqList([s.reverse() for s in seqs], seqs.alphabet)
if options.complement:
seqs = SeqList([Seq(s, seqs.alphabet).complement() for s in seqs], seqs.alphabet)
if isCodon:
if abs(options.codon_frame) > 1:
