aboutsummaryrefslogtreecommitdiff
path: root/wqflask/base/mrna_assay_tissue_data.py
diff options
context:
space:
mode:
authorPjotr Prins2016-06-17 06:48:01 +0000
committerPjotr Prins2016-06-17 06:48:01 +0000
commit93e3878c8b97ecbf28630e4bb3733a29f4cf45aa (patch)
treeef92de7c75b7899b406e99c97006a7e2b20bd86b /wqflask/base/mrna_assay_tissue_data.py
parenteb84f7c0e384e08b810e052fd3935f6d977b7ea2 (diff)
parent10df36b60273d81678f6630c07a2d8e5a6409282 (diff)
downloadgenenetwork2-93e3878c8b97ecbf28630e4bb3733a29f4cf45aa.tar.gz
Conflict
Diffstat (limited to 'wqflask/base/mrna_assay_tissue_data.py')
-rwxr-xr-xwqflask/base/mrna_assay_tissue_data.py42
1 files changed, 21 insertions, 21 deletions
diff --git a/wqflask/base/mrna_assay_tissue_data.py b/wqflask/base/mrna_assay_tissue_data.py
index ba82057a..039f5d1f 100755
--- a/wqflask/base/mrna_assay_tissue_data.py
+++ b/wqflask/base/mrna_assay_tissue_data.py
@@ -12,30 +12,30 @@ from MySQLdb import escape_string as escape
from pprint import pformat as pf
class MrnaAssayTissueData(object):
-
+
def __init__(self, gene_symbols=None):
self.gene_symbols = gene_symbols
self.have_data = False
if self.gene_symbols == None:
self.gene_symbols = []
-
+
#print("self.gene_symbols:", self.gene_symbols)
-
+
self.data = collections.defaultdict(Bunch)
-
+
#self.gene_id_dict ={}
#self.data_id_dict = {}
#self.chr_dict = {}
#self.mb_dict = {}
#self.desc_dict = {}
#self.probe_target_desc_dict = {}
-
+
query = '''select t.Symbol, t.GeneId, t.DataId, t.Chr, t.Mb, t.description, t.Probe_Target_Description
from (
select Symbol, max(Mean) as maxmean
from TissueProbeSetXRef
where TissueProbeSetFreezeId=1 and '''
-
+
# Note that inner join is necessary in this query to get distinct record in one symbol group
# with highest mean value
# Due to the limit size of TissueProbeSetFreezeId table in DB,
@@ -43,11 +43,11 @@ class MrnaAssayTissueData(object):
if len(gene_symbols) == 0:
query += '''Symbol!='' and Symbol Is Not Null group by Symbol)
as x inner join TissueProbeSetXRef as t on t.Symbol = x.Symbol
- and t.Mean = x.maxmean;
+ and t.Mean = x.maxmean;
'''
else:
in_clause = db_tools.create_in_clause(gene_symbols)
-
+
#ZS: This was in the query, not sure why: http://docs.python.org/2/library/string.html?highlight=lower#string.lower
query += ''' Symbol in {} group by Symbol)
@@ -56,7 +56,7 @@ class MrnaAssayTissueData(object):
'''.format(in_clause)
results = g.db.execute(query).fetchall()
-
+
lower_symbols = []
for gene_symbol in gene_symbols:
if gene_symbol != None:
@@ -68,7 +68,7 @@ class MrnaAssayTissueData(object):
if symbol.lower() in lower_symbols:
#gene_symbols.append(symbol)
symbol = symbol.lower()
-
+
self.data[symbol].gene_id = result.GeneId
self.data[symbol].data_id = result.DataId
self.data[symbol].chr = result.Chr
@@ -85,21 +85,21 @@ class MrnaAssayTissueData(object):
#function: get one dictionary whose key is gene symbol and value is tissue expression data (list type).
#Attention! All keys are lower case!
###########################################################################
-
+
def get_symbol_values_pairs(self):
id_list = [self.data[symbol].data_id for symbol in self.data]
print("id_list:", id_list)
symbol_values_dict = {}
-
+
query = """SELECT TissueProbeSetXRef.Symbol, TissueProbeSetData.value
FROM TissueProbeSetXRef, TissueProbeSetData
WHERE TissueProbeSetData.Id IN {} and
TissueProbeSetXRef.DataId = TissueProbeSetData.Id""".format(db_tools.create_in_clause(id_list))
-
+
print("TISSUE QUERY:", query)
-
+
results = g.db.execute(query).fetchall()
for result in results:
if result.Symbol.lower() not in symbol_values_dict:
@@ -110,11 +110,11 @@ class MrnaAssayTissueData(object):
#for symbol in self.data:
# data_id = self.data[symbol].data_id
# symbol_values_dict[symbol] = self.get_tissue_values(data_id)
-
-
+
+
return symbol_values_dict
-
-
+
+
#def get_tissue_values(self, data_id):
# """Gets the tissue values for a particular gene"""
#
@@ -133,7 +133,7 @@ class MrnaAssayTissueData(object):
# # symbol_values_pairs[symbol] = None
#
# return tissue_values
-
+
########################################################################################################
#input: cursor, symbolList (list), dataIdDict(Dict): key is symbol
#output: SymbolValuePairDict(dictionary):one dictionary of Symbol and Value Pair.
@@ -154,7 +154,7 @@ class MrnaAssayTissueData(object):
# #descDict,
# #pTargetDescDict = getTissueProbeSetXRefInfo(
# # GeneNameLst=GeneNameLst,TissueProbeSetFreezeId=TissueProbeSetFreezeId)
-#
+#
# if len(tissue_data.gene_symbols):
# return get_symbol_values_pairs(tissue_data)
-
+