def sliding_windows(rsaValues, win_width):
""" compute the average of sliding windows; return min,max"""
rsaseq = rsaValues
win_num = len(rsaseq) - win_width + 1
if win_num > 0:
#set the intial number
max_score = my_average(rsaseq[0:win_width])
min_score = max_score
for i in range(win_num):
onewin = rsaseq[i:i + win_width]
oneave = my_average(onewin)
# find the maximum
if oneave > max_score:
max_score = oneave
if oneave < min_score:
min_score = oneave
else: # win_num =0; or win_width larger than rsaValues
max_score = 0
min_score = 0
print("### rsaValues len: %d < win-width %d" %
(len(rsaValues), len(win_width)))
return (min_score, max_score)
def api_dowith_spinex(output_f, proseq, CUTOFF=0.25):
"""input fasta,folder,cutoff; ouput >id;seq;ss;exposed/burid"""
AA, SS = "", ""
ASA = []
RSA = []
BdEd = ""
if os.path.isfile(output_f):
AA, SS, ASA = get_asa_spinex(output_f)
else:
print("Can not find SpineX prediction of %s" % output_f)
#print(ASA)
if proseq != AA:
print("predicted vs. proseq different: %s" % output_f)
# do with the sequence contain X
#write the id;proseq;
tmp_seq = proseq.replace("X", "")
tmp_RSA, tmp_BdEd = ASA2RSA(tmp_seq, ASA)
#used the average rsa to replace the position x.
tmp_ave_RSA = my_average(tmp_RSA)
#write the buried/exposed
count_x = 0
for k in range(len(proseq)):
if proseq[k] == "X":
count_x = count_x + 1
if tmp_ave_RSA > CUTOFF:
BdEd += "1" #fw.write("1")
else:
BdEd += "0" #fw.write("0")
else:
if tmp_RSA[k - count_x] > CUTOFF:
BdEd += "1" #fw.write("1")
else:
BdEd += "0" #fw.write("0")
print("Meet %d X" % count_x)
# write the rsa values
for k in range(len(proseq)):
if proseq[k] == "X":
RSA.append(tmp_ave_RSA)
else:
RSA.append(tmp_RSA[k - count_x])
else:
RSA, BdEd = ASA2RSA(AA, ASA)
return ((AA, SS, BdEd, RSA))
def my_dowith_spinex(fr_n, folder, fw_n, CUTOFF=0.25):
"""input fasta,folder,fw_n, cutoff; ouput >id;seq;ss;exposed/burid"""
fr = open(fr_n, 'r')
lines = [line.strip() for line in fr.readlines()]
fr.close()
AA = ""
SS = ""
ASA = RSA = []
#fw_n="d62.rsa.dat" RSA values
fw = open(fw_n, 'w')
for i in range(len(lines)):
if len(lines[i]) > 0 and lines[i][0] == '>':
proid = lines[i].split()[0][1:]
proseq = lines[i + 1]
output_f = folder + "/" + proid + ".spXout" #"./spXout/%s.spXout" %proid.upper()
AA, SS, ASA = get_asa_spinex(output_f)
#print(ASA)
if proseq != AA:
print("predicted vs. proseq different: %s" % proid)
# do with the sequence contain X
#write the id;proseq;
fw.write("%s\n%s\n%s\n" % (lines[i], lines[i + 1], SS))
tmp_seq = proseq.replace("X", "")
tmp_RSA, tmp_BdEd = ASA2RSA(tmp_seq, ASA)
#used the average rsa to replace the position x.
tmp_ave_RSA = my_average(tmp_RSA)
#write the buried/exposed
count_x = 0
for k in range(len(proseq)):
if proseq[k] == "X":
count_x = count_x + 1
if tmp_ave_RSA > CUTOFF:
fw.write("1")
else:
fw.write("0")
else:
if tmp_RSA[k - count_x] > CUTOFF:
fw.write("1")
else:
fw.write("0")
fw.write("\n")
# write the rsa values
count_x = 0
for k in range(len(proseq)):
if proseq[k] == "X":
fw.write("%.5f," % tmp_ave_RSA)
count_x = count_x + 1
else:
fw.write("%.5f," % tmp_RSA[k - count_x])
fw.write("\n")
else:
RSA, BdEd = ASA2RSA(AA, ASA)
#write the id; proseq;ss,rsa_binary
fw.write("%s\n%s\n%s\n%s\n" % (lines[i], AA, SS, BdEd))
# write the rsa values
for j in range(len(RSA)):
fw.write("%.5f," % RSA[j])
fw.write("\n")
fw.close()
def my_dowith_spinex(fr_n, folder, CUTOFF=0.25):
"""input fasta,folder,cutoff; ouput >id;seq;ss;exposed/burid"""
fr = open(fr_n, 'r')
lines = [line.strip() for line in fr.readlines()]
fr.close()
SSS = []
BdEdS = []
RSAS = []
for i in range(len(lines)):
if len(lines[i]) > 0 and lines[i][0] == '>':
proid = lines[i].split()[0][1:]
proseq = lines[i + 1]
BdEd = ""
RSA = []
#output_f=folder+ "/"+proid+".spXout" #"./spXout/%s.spXout" %proid.upper()
output_f = "%s/%s.spXout" % (folder, proid)
if os.path.isfile(output_f):
AA, SS, ASA = get_asa_spinex(output_f)
else:
AA, SS = "", ""
ASA = []
print("Can not find SpineX prediction of %s" % proid)
#print(ASA)
if proseq != AA:
print("predicted vs. proseq different: %s" % proid)
# do with the sequence contain X
#write the id;proseq;
tmp_seq = proseq.replace("X", "")
tmp_RSA, tmp_BdEd = ASA2RSA(tmp_seq, ASA)
#used the average rsa to replace the position x.
tmp_ave_RSA = my_average(tmp_RSA)
#write the buried/exposed
count_x = 0
for k in range(len(proseq)):
if proseq[k] == "X":
count_x = count_x + 1
if tmp_ave_RSA > CUTOFF:
BdEd += "1" #fw.write("1")
else:
BdEd += "0" #fw.write("0")
else:
if tmp_RSA[k - count_x] > CUTOFF:
BdEd += "1" #fw.write("1")
else:
BdEd += "0" #fw.write("0")
print("Meet %d X" % count_x)
# write the rsa values
for k in range(len(proseq)):
if proseq[k] == "X":
RSA.append(tmp_ave_RSA)
else:
RSA.append(tmp_RSA[k - count_x])
else:
RSA, BdEd = ASA2RSA(AA, ASA)
SSS.append(SS)
BdEdS.append(BdEd)
RSAS.append(RSA)
return ((AA, SS, BdEdS, RSAS))