from Bio import motifs

from Bio.Alphabet import IUPAC

from Bio import Seq

#

# Download the most up-to-date version of the Jaspar nonredundant pfms.

import urllib

urllib.urlretrieve ("http://jaspar.genereg.net/html/DOWNLOAD/JASPAR_CORE/pfm/nonredundant/pfm_all.txt", "pfm_all.txt")

# There is a bug in Bio::motifs that throws an error if there are any blank lines in the jaspar database.

# This is a workaround to get rid of those lines in the downloaded pfm file.

#

fixed_pfm_file = open("pfm_all.fixed.txt", "w")

with open("pfm_all.txt") as f:

for line in f.readlines():

if line.strip():

fixed_pfm_file.write(line)

fixed_pfm_file.close()

# Output is printed to stdout to enable pipes to other process, etc. The output gives the effect of each possible mutation

# for each jaspar pfm. The file is tab delimited, with a header to make for easy reading into R or other downstream analyses.

# Each line of the output has the following fields:

#

# 1) name -- the name of the pfm in JASPAR

# 2) pos -- the relative position within the matrix. The value is from -1 to 1, where 0 is the center of the motif.

# 3-8) NN -- the change in pssm score associated with each possible mutation at that position in the motif.

#

print "name\tpos\tIC\tDegCons\tAG\tCT\tAC\tAT\tCG\tGT"

with open("pfm_all.fixed.txt") as handle:

for m in motifs.parse(handle, "jaspar"):

#

# Get the counts and the consensus motif for the pfm

#

counts = m.counts

cons = m.consensus

deg_cons = m.degenerate_consensus

#

# convert to pssm, adding a pseudocount of 0.1 to each base.

#

pssm = m.counts.normalize(pseudocounts=0.1).log_odds()

cons_score = pssm.calculate(cons)

cons_list = list(cons)

cons_str = str(cons)

deg_cons_str = str(deg_cons)

#

# for each position, generate a new test sequence for each possible nucleotide

# at that position. Then score that test sequence relative to the original pssm.

# Next, evaluate the absolute value of the score difference between every pair of

# test sequence and classify each pair as either a transition or transversion.

#

for i, c in enumerate(cons_list):

new_cons_str_A = Seq.Seq("".join((cons_str[0:i], "A", cons_str[i+1:])), IUPAC.unambiguous_dna)

new_cons_str_C = Seq.Seq("".join((cons_str[0:i], "C", cons_str[i+1:])), IUPAC.unambiguous_dna)

new_cons_str_G = Seq.Seq("".join((cons_str[0:i], "G", cons_str[i+1:])), IUPAC.unambiguous_dna)

new_cons_str_T = Seq.Seq("".join((cons_str[0:i], "T", cons_str[i+1:])), IUPAC.unambiguous_dna)

new_score_A = pssm.calculate(new_cons_str_A)

new_score_C = pssm.calculate(new_cons_str_C)

new_score_G = pssm.calculate(new_cons_str_G)

new_score_T = pssm.calculate(new_cons_str_T)

central_distance = 2 * (0.5 - float(i)/len(counts[1,:]))

pssm_position_score = pssm['A',i] + pssm['C',i] + pssm['G',i] + pssm['T',i]

print "%(name)s\t%(pos)f\t%(IC)f\t%(DegCons)s\t%(AG)f\t%(CT)f\t%(AC)f\t%(AT)f\t%(CG)f\t%(GT)f" % \

{'name': m.name, 'pos': central_distance, \

'IC': pssm_position_score, \

'DegCons': deg_cons_str[i], \

'AG': abs(new_score_A-new_score_G), \

'CT': abs(new_score_C-new_score_T), \

'AC': abs(new_score_A-new_score_C), \

'AT': abs(new_score_A-new_score_T), \

'CG': abs(new_score_C-new_score_G), \

'GT': abs(new_score_G-new_score_T)}