def run(self):
if self.prepare():
nTerrorT("Failed to prepare")
return True
# return
relationNames = glob("*.csv")
relationNames = [ relationName[:-4] for relationName in relationNames]
dbms = DBMS()
dbms.readCsvRelationList(relationNames, '.')
# This will overwrite the just read newMany2OneTable
newMany2OneTable = Relation('newMany2OneTable', dbms, columnList=['pdb_id', 'bmrb_id'])
bmrbIdNewMany2OneList = newMany2OneTable.getColumn('bmrb_id')
pdbIdNewMany2OneList = newMany2OneTable.getColumn('pdb_id')
tableScore_many2one = dbms.tables['score_many2one']
bmrbIdOldMany2OneList = tableScore_many2one.getColumn('bmrb_id')
pdbIdOldMany2OneList = tableScore_many2one.getColumn('pdb_id')
# NB that this table is a one 2 many unlike what I need.
tableAdit = dbms.tables['adit_nmr_matched_pdb_bmrb_entry_ids']
bmrbIdAditList = tableAdit.getColumn('bmrb_id')
pdbIdAditList = tableAdit.getColumn('pdb_id')
pdbIdAditList = [x.lower() for x in pdbIdAditList]
pdbIdAditNmrHash = list2dict( pdbIdAditList )
# New table from Dmitri with 655 matches.
# With corrections for invalids:
#15591,2k0b,1,SOLUTION NMR, 15591 should be matched to 2jy7
tableAdit2 = dbms.tables['BMRB_PDB_match']
bmrbIdAdit2List = tableAdit2.getColumn('BMRB_ID')
pdbIdAdit2List = tableAdit2.getColumn('PDB_ID')
pdbIdAdit2NmrHash = list2dict( pdbIdAdit2List )
# Manual corrections to Dmitri's table etc.
tableManual = dbms.tables['manualMatches'] # Maintain this list in SVN control.
bmrbIdManualList = tableManual.getColumn('bmrb_id')
pdbIdManualList = tableManual.getColumn('pdb_id')
_pdbIdManualNmrHash = list2dict( pdbIdManualList )
tablePdbNmrTable = dbms.tables['pdbNmrTable']
pdbIdPdbNmrList = tablePdbNmrTable.getColumn('pdb_id')
pdbIdPdbNmrHash = list2dict( pdbIdPdbNmrList )
bmrbTable = dbms.tables['bmrb']
bmrbIdList = bmrbTable.getColumn('bmrb_id')
bmrbIdHash = list2dict( bmrbIdList )
pdbIdListAbsent =[]
bmrbIdListAbsent =[]
for idx, pdb_id in enumerate(pdbIdOldMany2OneList):
bmrb_id = bmrbIdOldMany2OneList[idx]
if not pdbIdPdbNmrHash.has_key(pdb_id):
pdbIdListAbsent.append(pdb_id)
continue
if not bmrbIdHash.has_key(bmrb_id):
bmrbIdListAbsent.append(bmrb_id)
continue
bmrbIdNewMany2OneList.append(bmrb_id)
pdbIdNewMany2OneList.append(pdb_id)
l1 = len(pdbIdNewMany2OneList)
nTmessage("Skipped: %s obsolete PDB entries from score_many2one %s" % (len(pdbIdListAbsent),str(pdbIdListAbsent)))
nTmessage("Skipped: %s obsolete BMRB entries from score_many2one %s" % (len(bmrbIdListAbsent),str(bmrbIdListAbsent)))
nTmessage("Accepted from old list %s matches" % l1)
# Do both adit lists
pdbIdLoLDouble = [[],[]]
pdbIdLoLObsolete = [[],[]]
bmrbIdLoLObsolete =[[],[]]
ltotal1 = [ -1, -1]
l2 = [ -1, -1]
nadit = 2
for aditIdx, pdbIdAditXList in enumerate( [pdbIdAditList, pdbIdAdit2List] ):
bmrbIdAditXList = (bmrbIdAditList, bmrbIdAdit2List )[aditIdx]
pdbIdAditXNmrHash = (pdbIdAditNmrHash, pdbIdAdit2NmrHash )[aditIdx]
pdbIdListDouble = pdbIdLoLDouble[aditIdx]
pdbIdListObsolete = pdbIdLoLObsolete[aditIdx]
bmrbIdListObsolete = bmrbIdLoLObsolete[aditIdx]
for idx, pdb_id in enumerate(pdbIdAditXList):
bmrb_id = bmrbIdAditXList[idx]
if pdbIdAditXNmrHash[pdb_id] > 1:
if pdb_id not in pdbIdListDouble:
pdbIdListDouble.append(pdb_id)
continue
if not pdbIdPdbNmrHash.has_key(pdb_id):
pdbIdListObsolete.append(pdb_id)
continue
if not bmrbIdHash.has_key(bmrb_id):
bmrbIdListObsolete.append(bmrb_id)
continue
if pdb_id in pdbIdNewMany2OneList:
continue
# if bmrb_id in bmrbIdNewMany2OneList: allow this.
# continue
bmrbIdNewMany2OneList.append(bmrb_id)
pdbIdNewMany2OneList.append(pdb_id)
ltotal1[aditIdx] = len(pdbIdNewMany2OneList)
if aditIdx == 0:
l2[aditIdx] = ltotal1[aditIdx] - l1
else:
l2[aditIdx] = ltotal1[aditIdx] - ltotal1[aditIdx-1]
for idx, pdb_id in enumerate(pdbIdManualList):
bmrb_id = bmrbIdManualList[idx]
if not pdbIdPdbNmrHash.has_key(pdb_id):
nTerror("Failed to find %s in PDB; update the manual list." % pdb_id)
continue
if not bmrbIdHash.has_key(bmrb_id):
nTerror("Failed to find %s in BMRB; update the manual list." % bmrb_id)
continue
if pdb_id in pdbIdNewMany2OneList:
idx = pdbIdNewMany2OneList.index(pdb_id)
bmrb_id_current = bmrbIdNewMany2OneList[idx]
if bmrb_id_current == bmrb_id:
nTmessage("Already found %s in PDB with BMRB %s in manual and current list; consider updating the manual list." % (
pdb_id, bmrb_id))
continue
nTmessage("Using manual mapping of %s in PDB with BMRB %s in manual list instead of BMRB %s in current list." % (
pdb_id, bmrb_id, bmrb_id_current))
nTmessage("First removing match at idx %s in current list." % idx)
del bmrbIdNewMany2OneList[idx]
del pdbIdNewMany2OneList[idx]
# if bmrb_id in bmrbIdNewMany2OneList: allow this.
# continue
bmrbIdNewMany2OneList.append(bmrb_id)
pdbIdNewMany2OneList.append(pdb_id)
ltotal2 = len(pdbIdNewMany2OneList)
l3 = ltotal2 - ltotal1[nadit-1]
pdbIdNewHash = list2dict( pdbIdNewMany2OneList )
bmrbIdNewHash = list2dict( bmrbIdNewMany2OneList )
uniquePdbCount = len(pdbIdNewHash)
uniqueBmrbCount = len(bmrbIdNewHash)
nTmessage("Skipped: %s double entries from pdbIdAditList %s" % (len(pdbIdListDouble),str(pdbIdListDouble)))
for aditIdx in range(nadit):
pdbIdLoLObsolete[aditIdx].sort()
bmrbIdLoLObsolete[aditIdx].sort()
nTmessage("Skipped: %s obsolete PDB entries from adit%s %s" % (len( pdbIdLoLObsolete[aditIdx]), aditIdx,
str(pdbIdLoLObsolete[aditIdx])))
nTmessage("Skipped: %s obsolete BMRB entries from adit%s %s" % (len(bmrbIdLoLObsolete[aditIdx]), aditIdx,
str(bmrbIdLoLObsolete[aditIdx])))
nTmessage("Accepted from adit%s %s for a total of %s matches" %(aditIdx, l2[aditIdx], ltotal1[aditIdx]))
nTmessage("Accepted from manual list %s for a total of %s matches" %( l3, ltotal2))
nTmessage("Accepted unique %d PDB and %d BMRB entries" %( uniquePdbCount, uniqueBmrbCount))
pdbIdNewMany2OneNTList = NTlist(*pdbIdNewMany2OneList)
pdbIdDuplicateList = pdbIdNewMany2OneNTList.removeDuplicates()
if pdbIdDuplicateList:
nTerror("Got %s duplicate PDB entries in result: %s" % (len(pdbIdDuplicateList), str(pdbIdDuplicateList) ))
return True
bmrbIdNewMany2OneNTList = NTlist(*bmrbIdNewMany2OneList)
bmrbIdDuplicateList = bmrbIdNewMany2OneNTList.removeDuplicates()
bmrbIdDuplicateList = bmrbIdDuplicateList.removeDuplicates()
if bmrbIdDuplicateList:
nTmessage("Using %s BMRB entries that match two or more PDB entries." % len(bmrbIdDuplicateList) )
if newMany2OneTable.sortRelationByColumnIdx([0,1]):
nTerror("Failed to sort table: %s")
return True
newMany2OneTable.writeCsvFile()
def run(self):
if self.prepare():
nTerrorT("Failed to prepare")
return True
# return
relationNames = glob("*.csv")
relationNames = [relationName[:-4] for relationName in relationNames]
dbms = DBMS()
dbms.readCsvRelationList(relationNames, '.')
# This will overwrite the just read newMany2OneTable
newMany2OneTable = Relation('newMany2OneTable',
dbms,
columnList=['pdb_id', 'bmrb_id'])
bmrbIdNewMany2OneList = newMany2OneTable.getColumn('bmrb_id')
pdbIdNewMany2OneList = newMany2OneTable.getColumn('pdb_id')
tableScore_many2one = dbms.tables['score_many2one']
bmrbIdOldMany2OneList = tableScore_many2one.getColumn('bmrb_id')
pdbIdOldMany2OneList = tableScore_many2one.getColumn('pdb_id')
# NB that this table is a one 2 many unlike what I need.
tableAdit = dbms.tables['adit_nmr_matched_pdb_bmrb_entry_ids']
bmrbIdAditList = tableAdit.getColumn('bmrb_id')
pdbIdAditList = tableAdit.getColumn('pdb_id')
pdbIdAditList = [x.lower() for x in pdbIdAditList]
pdbIdAditNmrHash = list2dict(pdbIdAditList)
# New table from Dmitri with 655 matches.
# With corrections for invalids:
#15591,2k0b,1,SOLUTION NMR, 15591 should be matched to 2jy7
tableAdit2 = dbms.tables['BMRB_PDB_match']
bmrbIdAdit2List = tableAdit2.getColumn('BMRB_ID')
pdbIdAdit2List = tableAdit2.getColumn('PDB_ID')
pdbIdAdit2NmrHash = list2dict(pdbIdAdit2List)
# Manual corrections to Dmitri's table etc.
tableManual = dbms.tables[
'manualMatches'] # Maintain this list in SVN control.
bmrbIdManualList = tableManual.getColumn('bmrb_id')
pdbIdManualList = tableManual.getColumn('pdb_id')
_pdbIdManualNmrHash = list2dict(pdbIdManualList)
tablePdbNmrTable = dbms.tables['pdbNmrTable']
pdbIdPdbNmrList = tablePdbNmrTable.getColumn('pdb_id')
pdbIdPdbNmrHash = list2dict(pdbIdPdbNmrList)
bmrbTable = dbms.tables['bmrb']
bmrbIdList = bmrbTable.getColumn('bmrb_id')
bmrbIdHash = list2dict(bmrbIdList)
pdbIdListAbsent = []
bmrbIdListAbsent = []
for idx, pdb_id in enumerate(pdbIdOldMany2OneList):
bmrb_id = bmrbIdOldMany2OneList[idx]
if not pdbIdPdbNmrHash.has_key(pdb_id):
pdbIdListAbsent.append(pdb_id)
continue
if not bmrbIdHash.has_key(bmrb_id):
bmrbIdListAbsent.append(bmrb_id)
continue
bmrbIdNewMany2OneList.append(bmrb_id)
pdbIdNewMany2OneList.append(pdb_id)
l1 = len(pdbIdNewMany2OneList)
nTmessage("Skipped: %s obsolete PDB entries from score_many2one %s" %
(len(pdbIdListAbsent), str(pdbIdListAbsent)))
nTmessage("Skipped: %s obsolete BMRB entries from score_many2one %s" %
(len(bmrbIdListAbsent), str(bmrbIdListAbsent)))
nTmessage("Accepted from old list %s matches" % l1)
# Do both adit lists
pdbIdLoLDouble = [[], []]
pdbIdLoLObsolete = [[], []]
bmrbIdLoLObsolete = [[], []]
ltotal1 = [-1, -1]
l2 = [-1, -1]
nadit = 2
for aditIdx, pdbIdAditXList in enumerate(
[pdbIdAditList, pdbIdAdit2List]):
bmrbIdAditXList = (bmrbIdAditList, bmrbIdAdit2List)[aditIdx]
pdbIdAditXNmrHash = (pdbIdAditNmrHash, pdbIdAdit2NmrHash)[aditIdx]
pdbIdListDouble = pdbIdLoLDouble[aditIdx]
pdbIdListObsolete = pdbIdLoLObsolete[aditIdx]
bmrbIdListObsolete = bmrbIdLoLObsolete[aditIdx]
for idx, pdb_id in enumerate(pdbIdAditXList):
bmrb_id = bmrbIdAditXList[idx]
if pdbIdAditXNmrHash[pdb_id] > 1:
if pdb_id not in pdbIdListDouble:
pdbIdListDouble.append(pdb_id)
continue
if not pdbIdPdbNmrHash.has_key(pdb_id):
pdbIdListObsolete.append(pdb_id)
continue
if not bmrbIdHash.has_key(bmrb_id):
bmrbIdListObsolete.append(bmrb_id)
continue
if pdb_id in pdbIdNewMany2OneList:
continue
# if bmrb_id in bmrbIdNewMany2OneList: allow this.
# continue
bmrbIdNewMany2OneList.append(bmrb_id)
pdbIdNewMany2OneList.append(pdb_id)
ltotal1[aditIdx] = len(pdbIdNewMany2OneList)
if aditIdx == 0:
l2[aditIdx] = ltotal1[aditIdx] - l1
else:
l2[aditIdx] = ltotal1[aditIdx] - ltotal1[aditIdx - 1]
for idx, pdb_id in enumerate(pdbIdManualList):
bmrb_id = bmrbIdManualList[idx]
if not pdbIdPdbNmrHash.has_key(pdb_id):
nTerror("Failed to find %s in PDB; update the manual list." %
pdb_id)
continue
if not bmrbIdHash.has_key(bmrb_id):
nTerror("Failed to find %s in BMRB; update the manual list." %
bmrb_id)
continue
if pdb_id in pdbIdNewMany2OneList:
idx = pdbIdNewMany2OneList.index(pdb_id)
bmrb_id_current = bmrbIdNewMany2OneList[idx]
if bmrb_id_current == bmrb_id:
nTmessage(
"Already found %s in PDB with BMRB %s in manual and current list; consider updating the manual list."
% (pdb_id, bmrb_id))
continue
nTmessage(
"Using manual mapping of %s in PDB with BMRB %s in manual list instead of BMRB %s in current list."
% (pdb_id, bmrb_id, bmrb_id_current))
nTmessage("First removing match at idx %s in current list." %
idx)
del bmrbIdNewMany2OneList[idx]
del pdbIdNewMany2OneList[idx]
# if bmrb_id in bmrbIdNewMany2OneList: allow this.
# continue
bmrbIdNewMany2OneList.append(bmrb_id)
pdbIdNewMany2OneList.append(pdb_id)
ltotal2 = len(pdbIdNewMany2OneList)
l3 = ltotal2 - ltotal1[nadit - 1]
pdbIdNewHash = list2dict(pdbIdNewMany2OneList)
bmrbIdNewHash = list2dict(bmrbIdNewMany2OneList)
uniquePdbCount = len(pdbIdNewHash)
uniqueBmrbCount = len(bmrbIdNewHash)
nTmessage("Skipped: %s double entries from pdbIdAditList %s" %
(len(pdbIdListDouble), str(pdbIdListDouble)))
for aditIdx in range(nadit):
pdbIdLoLObsolete[aditIdx].sort()
bmrbIdLoLObsolete[aditIdx].sort()
nTmessage("Skipped: %s obsolete PDB entries from adit%s %s" %
(len(pdbIdLoLObsolete[aditIdx]), aditIdx,
str(pdbIdLoLObsolete[aditIdx])))
nTmessage("Skipped: %s obsolete BMRB entries from adit%s %s" %
(len(bmrbIdLoLObsolete[aditIdx]), aditIdx,
str(bmrbIdLoLObsolete[aditIdx])))
nTmessage("Accepted from adit%s %s for a total of %s matches" %
(aditIdx, l2[aditIdx], ltotal1[aditIdx]))
nTmessage("Accepted from manual list %s for a total of %s matches" %
(l3, ltotal2))
nTmessage("Accepted unique %d PDB and %d BMRB entries" %
(uniquePdbCount, uniqueBmrbCount))
pdbIdNewMany2OneNTList = NTlist(*pdbIdNewMany2OneList)
pdbIdDuplicateList = pdbIdNewMany2OneNTList.removeDuplicates()
if pdbIdDuplicateList:
nTerror("Got %s duplicate PDB entries in result: %s" %
(len(pdbIdDuplicateList), str(pdbIdDuplicateList)))
return True
bmrbIdNewMany2OneNTList = NTlist(*bmrbIdNewMany2OneList)
bmrbIdDuplicateList = bmrbIdNewMany2OneNTList.removeDuplicates()
bmrbIdDuplicateList = bmrbIdDuplicateList.removeDuplicates()
if bmrbIdDuplicateList:
nTmessage(
"Using %s BMRB entries that match two or more PDB entries." %
len(bmrbIdDuplicateList))
if newMany2OneTable.sortRelationByColumnIdx([0, 1]):
nTerror("Failed to sort table: %s")
return True
newMany2OneTable.writeCsvFile()
