def get_g_d():
gene_table = util.get_file_data("../final_data/gene.xls")
table = util.get_modified_data()
gene_list = []
mimnumber_result = []
gene_result = []
for i in range(gene_table.nrows):
if i == 0:
continue
else:
gene_list.append(
gene_table.row_values(i)[1] + gene_table.row_values(i)[2])
pattern = re.compile('[A-Za-z0-9]*-*[A-Za-z0-9]*,\s*\{[0-9]*\.?[0-9]*\}')
for i in range(table.nrows): # 循环逐行打印
if i == 0:
continue
else:
list = util.get_list_by_enter(table.row_values(i)[3]) # 获取基因数据
for item in list: # 对每一行数据进行处理
if item == '\n' or item == '':
continue
mimnumber_result.append(table.row_values(i)[0])
m = pattern.match(item)
if m != None:
temp = m.group()
gene = temp.split(',')[0]
code = temp.split(',')[1][temp.split(',')[1].find('{') +
1:temp.split(',')[1].find('}')]
gene_result.append(gene_list.index(gene + code) + 1)
else:
gene_result.append("")
wb = xlwt.Workbook()
sheet1 = wb.add_sheet('relationship')
sheet1.write(0, 0, 'id')
sheet1.write(0, 1, 'mimnumber')
sheet1.write(0, 2, 'gid')
for i in range(len(gene_result)):
sheet1.write(i + 1, 0, i + 1)
sheet1.write(i + 1, 1, mimnumber_result[i])
sheet1.write(i + 1, 2, gene_result[i])
wb.save("../final_data/g_d_relationship.xls")
def get_hpo():
current_date = util.get_current_date()
table = util.get_file_data('../final_data/system.xls')
src = []
dst = []
for i in range(table.nrows):
if i == 0:
continue
else:
src.append(table.row_values(i)[8])
for item in src:
if item=="":
continue
item = item[item.index(":") + 1:]
list = util.get_list_by_colon(item)
for l in list:
if l == "":
continue
else:
try:
dst.index(l)
except:
dst.append(l)
wb = xlwt.Workbook()
sheet1 = wb.add_sheet('sheet1')
sheet1.write(0, 0, 'hid')
sheet1.write(0, 1, 'hpo')
for i in range(len(dst)):
# print(result[i][j])
try:
sheet1.write(i + 1, 0, i + 1)
sheet1.write(i + 1, 1, dst[i])
except:
print(i, dst[i])
wb.save("../final_data/HPO.xls")
print("end")
def get_disease():
table = util.get_modified_data()
minnumber = []
title = []
inheritance = []
inheritance_dict = []
inheritance_table = util.get_file_data("../final_data/inheritance.xls")
for i in range(inheritance_table.nrows): # 循环逐行打印
if i == 0: # 跳过第一行
continue
else:
inheritance_dict.append(inheritance_table.row_values(i)[1])
for i in range(table.nrows): # 循环逐行打印
if i == 0: # 跳过第一行
continue
else:
minnumber.append(table.row_values(i)[0])
title.append(table.row_values(i)[1])
inheritance.append(table.row_values(i)[2])
prefered_title = []
shortening_title = []
for t in title:
index = 0
try:
index = t.index(';')
except:
index = len(t)
if index < len(t):
prefered_title.append(t[:index])
shortening_title.append(t[index + 2:])
else:
prefered_title.append(t)
shortening_title.append("null")
new_inheritance = []
for t in inheritance:
index = 0
try:
index = t.index(';')
except:
index = len(t)
if index < len(t):
new_inheritance.append(t[:index])
else:
new_inheritance.append(t)
wb = xlwt.Workbook()
sheet1 = wb.add_sheet('disease')
sheet1.write(0, 0, 'mimnumber')
sheet1.write(0, 1, 'preferredTitle')
sheet1.write(0, 2, 'shorteningTitle')
sheet1.write(0, 3, 'inheritance')
sheet1.write(0, 4, 'inheritance id')
for i in range(len(prefered_title)):
sheet1.write(i + 1, 0, minnumber[i])
sheet1.write(i + 1, 1, prefered_title[i])
sheet1.write(i + 1, 2, shortening_title[i])
sheet1.write(i + 1, 3, new_inheritance[i])
try:
sheet1.write(i + 1, 4,
inheritance_dict.index(new_inheritance[i]) + 1)
except:
sheet1.write(i + 1, 4, "")
wb.save("../final_data/disease.xls")