def pick_chunk_score(wdir, TAMO_file, target, genome):
'''Trims and returns the top motif in a cluster.
This script takes in the TAMO file from the motifs in a single cluster. It
trims the low-information ends from each motifs. It then indentifies the
motif that is most significantly represented in the target genes in your
genome. If no motif is significantly represented, then a blank top motif
file is created.
'''
os.system("cd %s" % wdir)
os.chdir(wdir)
script_dir = '/'.join(os.path.abspath(__file__).split('/')
[:-1]) # path to pcc_merge_CC.py script
##
# step 1 trim tamo to eliminate low information flanking sequence
trim_motif(TAMO_file, 0.1)
##
# step 2 Group Specificity Score" from the Church lab
# python MotifMetrics.py [Genes of interest] -genome [FASTA of promoter sequence] -t [Trimmed TAMO of cluster motifs]
# MotifMetrics.py checks if the motifs appear disproportionatly to the
# targets compared to the rest of the genes.
os.system(
"python %s/MotifMetrics.py %s -genome %s -t %s_0.1.trim -spec > %s_0.1.trim_Cout"
% (script_dir, target, genome, TAMO_file, TAMO_file))
##
# Gets the motif that is most significantly represented in your target genes
# Returns "None" if none of the motifs has a p-value above 0.001.
topm = parse_out_pcs("%s_0.1.trim_Cout" % TAMO_file)
print "topm", topm
##
# Writes the top motif to its own directory.
if topm != "None":
newdic = {}
ml = MotifTools.txt2motifs("%s_0.1.trim" % TAMO_file)
for m in ml:
if m.oneletter == topm:
newdic[m.oneletter] = m
save_motifs(newdic.values(), "%s.TOP" % TAMO_file)
os.system("rm %s_0.1.trim" % TAMO_file)
os.system("rm %s_0.1.trim_Cout" % TAMO_file)
##
# Writes a blank document if there was no top motif.
else:
oup = open("%s.TOP" % TAMO_file, "w")
oup.close()
def trim_motif(TAMO_file, cut=0.4):
'''Trims the motifs in TAMO_file, eliminating low-information flanks.'''
testmotifs = MotifTools.load(TAMO_file)
file = TAMO_file + "_" + str(cut) + ".trim"
new_mlist = []
for motif in testmotifs:
m = motif.trimmed(cut)
new_mlist.append(m)
save_motifs(new_mlist, file)
def TAMO_split(TAMO_file, motifs_per_file=190):
'''This function splits a TAMO into smaller files for create_cc'''
ml = MotifTools.txt2motifs(TAMO_file)
total = len(ml) / int(motifs_per_file) # Total number of TAMOs to generate
by = motifs_per_file
for i in range(total):
print i
print i * by + by, TAMO_file + '_n%s' % i
save_motifs(ml[i * by:i * by + by], TAMO_file + '_n%s' % i)
print total * by, len(ml), TAMO_file + '_n%s' % (total)
save_motifs(ml[total * by:len(ml)], TAMO_file + '_n%s' % (total))
return (total)
def tf2tamo(tf_file):
inp = open(tf_file, "r")
line = inp.readline()
motifs = []
while 1:
if not line:
break
if line.startswith("DE"):
name = line.strip().split()[1]
block = []
line = inp.readline()
while not line.startswith("XX"):
if not line:
break
block.append(line)
line = inp.readline()
motifs.append(parse_block(name, block))
else:
line = inp.readline()
save_motifs(motifs, '%s.tm' % (tf_file))
tamoMotifs = []
for jMat in jasparFiles:
tempMat = map(lambda l: l.strip().split(), open(jMat, "rU").readlines())
## eval() inteligently converts text numbers to int or float!
# for i in range(len(tempMat)):
# for j in range(len(tempMat[i])):
# tempMat[i][j] = eval(tempMat[i][j])
# transpose matrix
tempMat = transpose(mat(tempMat)).tolist()
for i in range(len(tempMat)):
tempMat[i] = {
"A": eval(tempMat[i][0]),
"C": eval(tempMat[i][1]),
"G": eval(tempMat[i][2]),
"T": eval(tempMat[i][3]),
}
jasparTAMO_Motif = Motif_from_counts(tempMat, bg=nucBack)
jasparTAMO_Motif.sourceFile = jMat.split("/")[-1]
tamoMotifs.append(jasparTAMO_Motif)
# print to file
cPickle.dump(tamoMotifs, open(pklFile, "w"))
save_motifs(tamoMotifs, tmoFile, kmer_count=60)
print "Done."
def merge_runs_cc(TAMO_file, wdir, height, distance, ancestor, target, genome):
'''This script is used to merge motifs with the PCC matrix of all motifs.
The script was originally written by Cheng Zou, and then converted to a
function by Alex Seddon.
'''
print "Here are the parameters you specified in this run "
print "-tamo %s" % TAMO_file
print "-wdir %s" % wdir
print "-h height to cut the tree, %s" % height
print "-ancestor %s" % ancestor
print "-target %s" % target
print "-genome %s" % genome
if TAMO_file == '' or wdir == '':
help()
os.system("cd %s" % wdir)
os.chdir(wdir)
# Get the directory where the script is located.
script_dir = '/'.join(os.path.abspath(__file__).split('/')[:-1])
# This code was in the original clustering script. It has been taken out
# because the processes involved take too long and have been taken up by
# the matrrix creation scripts and the run_UPGMA script.
#if distance==0:
# os.system("python /mnt/home/seddonal/gil scottscripts/5_motif_merging/3.calculate_distance_matrix.py -i %s --dfunc pccrange" % TAMO_file)
#os.system("R --vanilla --slave --args %s.dm %s< /mnt/home/seddonal/scripts/5_motif_merging/UPGMA_final.R> %s.Rout" % (TAMO_file,height,TAMO_file))
cl_dic = {}
n = 0
# The file, TAMO_file.dm_UPGMA_Cl_0.05, is inorder of the motifs that appear
# in the TAMO_file. If two motifs have the same number, they are considered
# a part of the same cluster.
# This loop pulls the clustering information out of this file and creats
# the dictionary cl_dic = {cluster_index:{motif_index:'1'}}
for line in open("%s.dm_UPGMA_Cl_%s" % (TAMO_file, height), "r"):
# Gets the clusterindex of this motif
cl = line.strip()
# Adds the cluster index if it has not been
if not cl_dic.has_key(cl):
cl_dic[cl] = {}
cl_dic[cl][n] = "1" # Adds the motif to that cluster
n += 1 # Increases the motif index for the next motif
#print cl_dic
ml = MotifTools.txt2motifs(TAMO_file)
old = [] # List of motifs that are the sole members of a cluster.
# I think I can divide up this portion of the code to create a series
print ancestor, ancestor == 0
cc_output = open('merge_runs_cc', 'w')
if ancestor == 0:
# This loop Looks at each cluster and attempts to merge the motifs
# in the cluster if there are multiple motifs.
for i in cl_dic.keys():
print i, cl_dic[i]
# If there are multiple motifs in the cluster, it merges the motifs
if len(cl_dic[i]) > 1:
# Adds all of the motifs in the cluster to an object called
# mlist.
mlist = []
for j in cl_dic[i]:
mlist.append(ml[j])
# Saves these motifs to there own TAMO file.
save_motifs(mlist, "%s_sub_%s.tm" % (TAMO_file, i))
cc_output.write(
'module load TAMO; python %s/pcc_merge_CC.py merge_runs_no_ancestor -t %s/%s -i %s -target %s -genome %s\n'
% (script_dir, wdir, TAMO_file, i, target, genome))
# If there is only one motif in the cluster, it leaves it alone,
# And adds it to old
else:
key = cl_dic[i].keys()[0]
old.append(ml[key])
if ancestor == 1:
# This loop Looks at each cluster and attempts to merge the motifs
# in the cluster if there are multiple motifs.
for i in cl_dic.keys():
print i, cl_dic[i]
# If there are multiple motifs in the cluster, it merges the motifs
if len(cl_dic[i]) > 1:
# Adds all of the motifs in the cluster to an object called
# mlist.
mlist = []
for j in cl_dic[i]:
mlist.append(ml[j])
# Saves these motifs to there own TAMO file.
save_motifs(mlist, "%s_sub_%s.tm" % (TAMO_file, i))
cc_output.write(
'module load TAMO; module load STAMPmotif; python %s/pcc_merge_CC.py merge_runs_ancestor -t %s/%s -i %s -target %s -genome %s\n'
% (script_dir, wdir, TAMO_file, i, target, genome))
else:
key = cl_dic[i].keys()[0]
old.append(ml[key])
# Combine together the motifs that are in there own cluster.
#os.system("cat %s_sub_*_sum.tm.tf.tm.TOP > %s_sub_new.tm" % (TAMO_file,TAMO_file))
save_motifs(old, "%s_sub_old.tm" % (TAMO_file))
def merge_runs(TAMO_file, wdir, height, distance, ancestor, target, genome):
'''This script is used to merge motifs with the PCC matrix of all motifs.
The script was originally written by Cheng Zou, and then converted to a
function by Alex Seddon.
'''
print "Here are the parameters you specified in this run "
print "-tamo %s" % TAMO_file
print "-wdir %s" % wdir
print "-h height to cut the tree, %s" % height
print "-distance %s" % distance
print "-ancestor %s" % ancestor
print "-target %s" % target
print "-genome %s" % genome
if TAMO_file == '' or wdir == '':
help()
os.system("cd %s" % wdir)
os.chdir(wdir)
# This code was in the original clustering script. It has been taken out
# because the processes involved take too long and have been replaced by
# the matrix creation scripts and the run_UPGMA script.
#if distance==0:
# os.system("python /mnt/home/seddonal/scripts/5_motif_merging/3.calculate_distance_matrix.py -i %s --dfunc pccrange" % TAMO_file)
#os.system("R --vanilla --slave --args %s.dm %s< /mnt/home/seddonal/scripts/5_motif_merging/UPGMA_final.R> %s.Rout" % (TAMO_file,height,TAMO_file))
cl_dic = {}
n = 0
# The file, TAMO_file.dm_UPGMA_Cl_0.05, is inorder of the motifs that appear
# in the TAMO_file. If two motifs have the same number, they are considered
# a part of the same cluster.
# This loop pulls the clustering information out of this file and creats
# the dictionary cl_dic = {cluster_index:{motif_index:'1'}}
for line in open("%s.dm_UPGMA_Cl_%s" % (TAMO_file, height), "r"):
# Gets the clusterindex of this motif
cl = line.strip()
# Adds the cluster index if it has not been
if not cl_dic.has_key(cl):
cl_dic[cl] = {}
cl_dic[cl][n] = "1" # Adds the motif to that cluster
n += 1 # Increases the motif index for the next motif
#print cl_dic
ml = MotifTools.txt2motifs(TAMO_file)
old = [] # List of motifs that are the sole members of a cluster.
# I think I can divide up this portion of the code to create a series
print ancestor, ancestor == 0
if ancestor == 0:
# This loop Looks at each cluster and attempts to merge the motifs
# in the cluster if there are multiple motifs.
for i in cl_dic.keys():
print i, cl_dic[i]
# If there are multiple motifs in the cluster, it merges the motifs
if len(cl_dic[i]) > 1:
# Adds all of the motifs in the cluster to an object called
# mlist.
mlist = []
for j in cl_dic[i]:
mlist.append(ml[j])
# Saves these motifs to there own TAMO file.
save_motifs(mlist, "%s_sub_%s.tm" % (TAMO_file, i))
# I am fairly certain that this process of converting to TF and
# then returning it to TAMO format is only for keeping the names
# consistent. I need to verify this suspicion
tamo2tf("%s_sub_%s.tm" % (TAMO_file, i))
os.system("cat %s_sub_%s.tm.tf > %s_sub_%s_sum.tm.tf" %
(TAMO_file, i, TAMO_file, i))
tf2tamo("%s_sub_%s_sum.tm.tf" % (TAMO_file, i))
# Gets the top motif in the cluster.
pick_chunk_score(wdir,
'%s_sub_%s_sum.tm.tf.tm' % (TAMO_file, i),
target, genome)
# Removes the files that were created.
os.system("rm %s_sub_%s_sum.tm.tf.tm" % (TAMO_file, i))
os.system("rm %s_sub_%s_sum.tm.tf" % (TAMO_file, i))
os.system("rm -R %s_sub_%s.tm.tf_ST*" % (TAMO_file, i))
# If there is only one motif in the cluster, it leaves it alone,
# And adds it to old
else:
key = cl_dic[i].keys()[0]
old.append(ml[key])
if ancestor == 1:
# This loop Looks at each cluster and attempts to merge the motifs
# in the cluster if there are multiple motifs.
for i in cl_dic.keys():
print i, cl_dic[i]
# If there are multiple motifs in the cluster, it merges the motifs
if len(cl_dic[i]) > 1:
# Adds all of the motifs in the cluster to an object called
# mlist.
mlist = []
for j in cl_dic[i]:
mlist.append(ml[j])
# Saves these motifs to there own TAMO file.
save_motifs(mlist, "%s_sub_%s.tm" % (TAMO_file, i))
# Merges the motifs in the same cluster using STAMP
tamo2tf("%s_sub_%s.tm" % (TAMO_file, i))
# Gets the JASPER motifs that best match the motifs from within
# the cluster.
os.system(
"STAMP -tf %s_sub_%s.tm.tf -sd /home/chengzou/bin/STAMP/ScoreDists/JaspRand_PCC_SWU.scores \
-go 1000 -ge 1000 -cc PCC -align SWU -out %s_sub_%s.tm.tf_STout -chp > %s_sub_%s.tm.tf_STout.log"
% (TAMO_file, i, TAMO_file, i, TAMO_file, i))
parse_out_STAMP(TAMO_file, i)
# combines the JASPER motifs with the cluster motif and then
# converts them all to one TAMO file
os.system(
"cat %s_sub_%s.tm.tf %s_sub_%s.tm.tf_SToutFBP.txt.mod %s_sub_%s.tm.tf_STout_tree_clusters.txt > %s_sub_%s_sum.tm.tf"
% (TAMO_file, i, TAMO_file, i, TAMO_file, i, TAMO_file, i))
tf2tamo("%s_sub_%s_sum.tm.tf" % (TAMO_file, i))
# Gets the top motif within the TAMO file.
pick_chunk_score(wdir,
'%s_sub_%s_sum.tm.tf.tm' % (TAMO_file, i),
target, genome)
# Removes any files created in the processing.
os.system("rm %s_sub_%s_sum.tm.tf.tm" % (TAMO_file, i))
os.system("rm %s_sub_%s_sum.tm.tf" % (TAMO_file, i))
os.system("rm -R %s_sub_%s.tm.tf_ST*" % (TAMO_file, i))
else:
key = cl_dic[i].keys()[0]
old.append(ml[key])
# Combine together the top motifs from every
os.system("cat %s_sub_*_sum.tm.tf.tm.TOP > %s_sub_new.tm" %
(TAMO_file, TAMO_file))
save_motifs(old, "%s_sub_old.tm" % (TAMO_file))
os.system("cat %s_sub_old.tm %s_sub_new.tm > %s_P1.tm" %
(TAMO_file, TAMO_file, TAMO_file))
tar_dir = sys.argv[i + 1].rstrip("/")
print tar_dir
ml = MotifTools.txt2motifs(file)
n = 0
new_list = []
##
# Looks at each motif from the TAMO file. Uses the find function from
# motility to find the sequences with that motif.
for Ikey in range(len(ml)):
#print m.ll
time1 = time.time()
m = ml[Ikey] # Pull out the motif from the motif list.
save_motifs([m], file + '_' + str(Ikey)) # Save the motif as a file.
##
# Create the PWM matrix from the TAMO motif.
matrix = []
for i in range(len(m.ll)):
position = []
for nt in ["A", "C", "G", "T"]:
position.append(m.ll[i][nt])
matrix.append(position)
#print matrix
pwm = motility.PWM(matrix)
#
##
##