def get_protein_sequence_descriptor(Uniprot_ID_list,
no_compitable_sequence_file_name):
#write a file has non-compatable protein sequence...
#text_file = open("non_compitable_sequence_uniprot.txt", "w")
#text_file = open("non_compitable_sequence_drugbank.txt", "w")
text_file = open(no_compitable_sequence_file_name, "w")
descr = []
error_protein_list = []
for protein in Uniprot_ID_list:
proseq = gps(protein)
Des = GetProDes(proseq)
print protein
try:
alldes = Des.GetALL()
PSD_list = []
for key in alldes:
PSD_list.append(alldes[key])
descr.append(PSD_list)
except (KeyError, IOError, AttributeError, ZeroDivisionError):
error_protein_list.append(protein)
text_file.write(protein)
text_file.write("\n")
pass
text_file.close()
descr = np.asarray(descr)
return descr
def test_p48039():
proteinsequence = gps("P48039") # download the protein sequence by uniprot id
DesObject = PyPro.GetProDes(proteinsequence) # construct a GetProDes object
print(DesObject.GetCTD()) # calculate 147 CTD descriptors
print(DesObject.GetAAComp()) # calculate 20 amino acid composition descriptors
paac = DesObject.GetPAAC(
lamda=10, weight=0.05
) # calculate 30 pseudo amino acid composition descriptors
for i in paac:
print(i)
def test_p33765():
# TODO: "P33765" gives "HTTP Error 300" (abiguity?) Why?
proteinsequence = gps(
"P33765") # download the protein sequence by uniprot id
DesObject = PyPro.GetProDes(
proteinsequence) # construct a GetProDes object
print(DesObject.GetCTD()) # calculate 147 CTD descriptors
print(DesObject.GetAAComp()
) # calculate 20 amino acid composition descriptors
paac = DesObject.GetPAAC(
lamda=10,
weight=0.05) # calculate 30 pseudo amino acid composition descriptors
for i in paac:
print(i)
def test_docs():
uniprotid = "P48039"
gps(uniprotid) # Check the return value!