def get_translation_from_GeneID(self, GeneID):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
return phamerator_manage_db.get_translation_from_GeneID(self.c, GeneID)
def get_scores(self, query, subject):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
return phamerator_manage_db.get_scores(self.c, query, subject)
def get_unique_phams(self):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
return phamerator_manage_db.get_unique_phams(self.c)
def create_genome_map(self,phages):
self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor()
self.db = pham.db(c = self.c)
lengths = []
for phage in phages:
length = phamerator_manage_db.get_length_of_genome(self.c,phage["PhageID"])
lengths.append(length)
length = max(lengths)
self.canvas = goocanvas.Canvas()
canvasInterface = cartographer.CanvasInterface(self.c, self.canvas)
self.canvas.set_root_item_model(canvasInterface.create_canvas_model(phages, length))
zoomFactor = 20.0
self.canvas.set_bounds(0,0,length/zoomFactor, 100000)
needs_new = True
while needs_new == True:
name = "/tmp/phageSOAP/tmp" + str(hash(str(random.randint(1,64563))+str(hash(self.canvas))))
if os.path.exists(name):
needs_new = True
else:
needs_new = False
self.current_genome_map = name
surface = cairo.SVGSurface (self.current_genome_map, (length/zoomFactor)+72, 5*72*len(phages))
cr = cairo.Context (surface)
cr.translate (36, 130)
self.canvas.render (cr, None, 0.1)
cr.show_page ()
print "\n>genome map generated\n"
return "genome map generated\n"
def get_genes_from_PhageID(self, PhageID):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
return phamerator_manage_db.get_genes_from_PhageID(self.c, PhageID)
def get_fasta_from_pham(self, phamName):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
return phamerator_manage_db.get_fasta_from_pham(self.c, phamName)
def get_all_scores(self, alignmentType='both'):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
return phamerator_manage_db.get_all_scores(self.c,
alignmentType=alignmentType)
def get_gene_start_stop_length_orientation_from_GeneID(self, GeneID):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
return phamerator_manage_db.get_gene_start_stop_length_orientation_from_GeneID(
self.c, GeneID)
def create_pham_circle(self,phamName,alignmentColor,adjustment,radius):
self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor()
self.db = pham.db(c = self.c)
filename = str(phamName) + str(alignmentColor) + str(adjustment) + str(radius)
filelist = os.listdir(self.cachedir)
for item in filelist:
if item == filename:
self.current_phamCircle = self.cachedir + filename
string = "phamCircle " + str(phamName) + " found in cache, using cached version\n"
print ">" + string
return string
self.radius = radius
self.alignmentColor = alignmentColor
self.threshold = 0.70
self.phamCircleCanvas = goocanvas.Canvas()
GeneIDs = phamerator_manage_db.get_members_of_pham(self.c, phamName)
memberPhages, self.nonMemberPhages = [], []
for GeneID in GeneIDs:
PhageID = phamerator_manage_db.get_PhageID_from_GeneID(self.c, GeneID)
if PhageID not in memberPhages: memberPhages.append(PhageID)
totalPhages = phamerator_manage_db.get_PhageIDs(self.c)
for p in totalPhages:
if p not in memberPhages: self.nonMemberPhages.append(phamerator_manage_db.get_phage_name_from_PhageID(self.c, p))
self.l = []
self.genes = []
for a in GeneIDs:
for b in GeneIDs:
if a != b:
for gene in [a, b]:
if gene not in self.genes: self.genes.append(gene)
clustalwScore, blastScore = phamerator_manage_db.get_scores(self.c, a, b)
if clustalwScore >= 0.275: self.l.append((a, b, 'clustalw',clustalwScore))
if blastScore and blastScore <= 0.0001: self.l.append((a, b, 'blast',blastScore))
self.phamCircle = PhamDisplay.PhamCircle(phamName, self.c,verbose=True,radius=self.radius)
if self.alignmentColor == True:
self.phamCircleCanvas.set_root_item_model(self.phamCircle.create_canvas_model(self.nonMemberPhages, self.genes, self.l,adjustment,self.threshold,blastColor='#ff0000', clustalwColor='#0000ff'))
else:
phamColorFromDataBase = self.db.select('pham_color','color',name = phamName)[0][0]
self.phamCircleCanvas.set_root_item_model(self.phamCircle.create_canvas_model(self.nonMemberPhages, self.genes, self.l,adjustment,self.threshold,allColor = phamColorFromDataBase))
"""x, y = (600, 500)
self.phamCircleCanvas.set_size_request(x, y)
self.defaultPhamCircleCanvasSize = (x, y)
self.phamCircleCanvas.show()
self.window.window.set_cursor(None)
return False"""
self.current_phamCircle = self.cachedir + filename
self.phamCircleCanvas.set_bounds(0,0, 10000, 10000)
surface = cairo.SVGSurface (self.current_phamCircle, 15*72, 15*72)
cr = cairo.Context (surface)
cr.translate (10, 0)
self.phamCircleCanvas.render (cr, None, 0.1)
cr.show_page()
string = "phamCircle " + str(phamName) + " generated\n"
print ">" + string
return string
def get_GeneIDs(self, type=None, PhageID=None):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
return phamerator_manage_db.get_GeneIDs(self.c,
type=type,
PhageID=PhageID)
def get_number_of_pham_members(self, phamName, PhageID=None):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
return phamerator_manage_db.get_number_of_pham_members(self.c,
phamName,
PhageID=PhageID)
def get_relatives(self,
GeneID,
alignmentType='both',
blastThreshold=None,
clustalwThreshold=None):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
return phamerator_manage_db.get_relatives(
self.c,
GeneID,
alignmentType=alignmentType,
blastThreshold=blastThreshold,
clustalwThreshold=clustalwThreshold)
def create_genome_map(self, phages):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
lengths = []
for phage in phages:
length = phamerator_manage_db.get_length_of_genome(
self.c, phage["PhageID"])
lengths.append(length)
length = max(lengths)
self.canvas = goocanvas.Canvas()
canvasInterface = cartographer.CanvasInterface(self.c, self.canvas)
self.canvas.set_root_item_model(
canvasInterface.create_canvas_model(phages, length))
zoomFactor = 20.0
self.canvas.set_bounds(0, 0, length / zoomFactor, 100000)
needs_new = True
while needs_new == True:
name = "/tmp/phageSOAP/tmp" + str(
hash(str(random.randint(1, 64563)) + str(hash(self.canvas))))
if os.path.exists(name):
needs_new = True
else:
needs_new = False
self.current_genome_map = name
surface = cairo.SVGSurface(self.current_genome_map,
(length / zoomFactor) + 72,
5 * 72 * len(phages))
cr = cairo.Context(surface)
cr.translate(36, 130)
self.canvas.render(cr, None, 0.1)
cr.show_page()
print "\n>genome map generated\n"
return "genome map generated\n"
def get_all_scores(self, alignmentType='both'): self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor() self.db = pham.db(c = self.c) return phamerator_manage_db.get_all_scores(self.c, alignmentType=alignmentType)
def get_genes_from_PhageID(self, PhageID): self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor() self.db = pham.db(c = self.c) return phamerator_manage_db.get_genes_from_PhageID(self.c, PhageID)
def get_GeneIDs(self, type=None, PhageID=None): self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor() self.db = pham.db(c = self.c) return phamerator_manage_db.get_GeneIDs(self.c, type=type, PhageID=PhageID)
def get_seq_from_GeneID(self, GeneID, extra=None): self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor() self.db = pham.db(c = self.c) return phamerator_manage_db.get_seq_from_GeneID(self.c, GeneID, extra=extra)
def main():
argDict = get_options(sys.argv[1:])
# cfg = ConfigParser.RawConfigParser()
# cfg.read(os.path.join(os.environ['HOME'], '.my.cnf'))
# try:
# username = cfg.get('client','user')
# except ConfigParser.NoOptionError:
if argDict['user']:
username = argDict['user']
else:
username = raw_input('database username: '******'client','password')
# except ConfigParser.NoOptionError:
if "password" in argDict:
password = getpass.getpass('database password: '******''
database = argDict['database']
server = argDict['server']
cthreshold = argDict['clustalw_threshold']
bthreshold = argDict['blast_threshold']
c = db_conf.db_conf(username=username,
password=password,
server=server,
db=database).get_cursor()
db = pham.db(c)
GeneIDs = get_GeneIDs(c)
# get all the phams that are in the database
oldController = pham.PhamController(c, source='db')
# create an empty PhamController object that will be populated with phams for pre-existing genes only
# this checks to make sure these phams are still valid based on new BLAST scores
currentController = pham.PhamController(c)
new_genes = []
for GeneID in GeneIDs:
if not oldController.find_phams_with_gene(str(GeneID)):
#print '%s is not in a pham' % str(GeneID)
new_genes.append(str(GeneID))
print 'there are %s genes that are not assigned to a pham' % len(new_genes)
print 'ignoring these and verifying the old phams...'
# Make sure that the existing phams in the database are still valid based on the
# current alignment scores
for GeneID in GeneIDs:
if GeneID in new_genes: continue
relatives = follow_rel_chain2(c,
GeneID, [],
ignore=new_genes,
cthreshold=cthreshold,
bthreshold=bthreshold)
p = pham.Pham(name=None, members=[], children=[])
p.add_members((GeneID, ))
p.add_members(relatives)
print 'created pham'
try:
#print 'creating a new pham and adding it to the currentController:', p
currentController.add_pham(p)
except pham.DuplicatePhamError:
pass
#print 'pham %s is already in the current controller:' % p
# Look for phams that should be joined because they each contain the same gene.
# Join any that are found.
for GeneID in GeneIDs:
pwg = currentController.find_phams_with_gene(
str(GeneID)) #pwg = phams with gene
#print 'GeneID: %s, pwg: %s' % (GeneID, pwg)
if pwg:
# one (and only one) pham already contains this gene, don't do anything
if len(pwg) == 1:
p = pwg[0]
# if this gene is already in more than one pham, join those phams
elif len(pwg) > 1:
p = currentController.join_phams(pwg)
print "there are %s total phams when ignoring new genes" % len(
currentController.phams)
currentController = currentController - oldController
print "there are %s split phams that need to be phixed" % len(
currentController.phams)
currentController.save()
# create an empty PhamController object that will be populated with phams for new and pre-existing genes
# the pre-existing phams will be subtracted out of this object's pham list
newController = pham.PhamController(c)
oldController = pham.PhamController(c, source='db')
# for every gene in the database, figure out if it is already in a pham
for GeneID in GeneIDs:
append = True
relatives = follow_rel_chain2(c,
GeneID, [],
cthreshold=cthreshold,
bthreshold=bthreshold)
pwg = newController.find_phams_with_gene(
str(GeneID)) #pwg = phams with gene
#print 'GeneID: %s, pwg: %s' % (GeneID, pwg)
if pwg:
# one (and only one) pham already contains this gene, don't do anything
if len(pwg) == 1:
p = pwg[0]
# if this gene is already in more than one pham, join those phams
elif len(pwg) > 1:
###print 'joining phams...'
#for item in pwg: print item.members
p = newController.join_phams(pwg)
# if this gene isn't in a pham already, create a new pham
else:
p = pham.Pham(name=None, members=[], children=[])
p.add_members((GeneID, ))
# add any genes related to this gene to the pham that it's in
p.add_members(relatives)
toJoin = [p]
for gene in p.members:
for result in newController.find_phams_with_gene(gene):
if result not in toJoin: toJoin.append(result)
if len(toJoin) > 1:
#print 'joining phams with these members:'
#for j in toJoin: print j, j.name, j.members
p = newController.join_phams(toJoin)
if append:
###print 'adding pham', p, 'with members:', p.members
try:
newController.add_pham(p)
except:
pham.DuplicatePhamError
newController = newController - oldController
print "there are", len(newController.phams), "new phams."
for newPham in newController.phams:
print newPham
print "Saving..."
newController.save()
def get_number_of_pham_members(self, phamName, PhageID=None): self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor() self.db = pham.db(c = self.c) return phamerator_manage_db.get_number_of_pham_members(self.c, phamName, PhageID=PhageID)
def get_unique_phams(self): self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor() self.db = pham.db(c = self.c) return phamerator_manage_db.get_unique_phams(self.c)
def get_pham(self):
db = pham.db(c)
pham = db.select('pham', name, GeneID = self.GeneID)
return pham
def main():
argDict = get_options(sys.argv[1:])
# cfg = ConfigParser.RawConfigParser()
# cfg.read(os.path.join(os.environ['HOME'], '.my.cnf'))
# try:
# username = cfg.get('client','user')
# except ConfigParser.NoOptionError:
if argDict['user']:
username = argDict['user']
else:
username = raw_input('database username: '******'client','password')
# except ConfigParser.NoOptionError:
if "password" in argDict:
password = getpass.getpass('database password: '******''
database = argDict['database']
server = argDict['server']
cthreshold = argDict['clustalw_threshold']
bthreshold = argDict['blast_threshold']
c = db_conf.db_conf(username=username, password=password, server=server, db=database).get_cursor()
db = pham.db(c)
GeneIDs = get_GeneIDs(c)
# get all the phams that are in the database
oldController = pham.PhamController(c, source='db')
# create an empty PhamController object that will be populated with phams for pre-existing genes only
# this checks to make sure these phams are still valid based on new BLAST scores
currentController = pham.PhamController(c)
new_genes = []
for GeneID in GeneIDs:
if not oldController.find_phams_with_gene(str(GeneID)):
#print '%s is not in a pham' % str(GeneID)
new_genes.append(str(GeneID))
print 'there are %s genes that are not assigned to a pham' % len(new_genes)
print 'ignoring these and verifying the old phams...'
# Make sure that the existing phams in the database are still valid based on the
# current alignment scores
for GeneID in GeneIDs:
if GeneID in new_genes: continue
relatives = follow_rel_chain2(c, GeneID, [], ignore=new_genes, cthreshold=cthreshold, bthreshold=bthreshold)
p = pham.Pham(name=None, members=[], children=[])
p.add_members((GeneID,))
p.add_members(relatives)
print 'created pham'
try:
#print 'creating a new pham and adding it to the currentController:', p
currentController.add_pham(p)
except pham.DuplicatePhamError:
pass
#print 'pham %s is already in the current controller:' % p
# Look for phams that should be joined because they each contain the same gene.
# Join any that are found.
for GeneID in GeneIDs:
pwg = currentController.find_phams_with_gene(str(GeneID)) #pwg = phams with gene
#print 'GeneID: %s, pwg: %s' % (GeneID, pwg)
if pwg:
# one (and only one) pham already contains this gene, don't do anything
if len(pwg) == 1:
p = pwg[0]
# if this gene is already in more than one pham, join those phams
elif len(pwg) > 1:
p = currentController.join_phams(pwg)
print "there are %s total phams when ignoring new genes" % len(currentController.phams)
currentController = currentController - oldController
print "there are %s split phams that need to be phixed" % len(currentController.phams)
currentController.save()
# create an empty PhamController object that will be populated with phams for new and pre-existing genes
# the pre-existing phams will be subtracted out of this object's pham list
newController = pham.PhamController(c)
oldController = pham.PhamController(c, source='db')
# for every gene in the database, figure out if it is already in a pham
for GeneID in GeneIDs:
append=True
relatives = follow_rel_chain2(c, GeneID, [], cthreshold=cthreshold, bthreshold=bthreshold)
pwg = newController.find_phams_with_gene(str(GeneID)) #pwg = phams with gene
#print 'GeneID: %s, pwg: %s' % (GeneID, pwg)
if pwg:
# one (and only one) pham already contains this gene, don't do anything
if len(pwg) == 1:
p = pwg[0]
# if this gene is already in more than one pham, join those phams
elif len(pwg) > 1:
###print 'joining phams...'
#for item in pwg: print item.members
p = newController.join_phams(pwg)
# if this gene isn't in a pham already, create a new pham
else:
p = pham.Pham(name=None, members=[], children=[])
p.add_members((GeneID,))
# add any genes related to this gene to the pham that it's in
p.add_members(relatives)
toJoin = [p]
for gene in p.members:
for result in newController.find_phams_with_gene(gene):
if result not in toJoin: toJoin.append(result)
if len(toJoin) > 1:
#print 'joining phams with these members:'
#for j in toJoin: print j, j.name, j.members
p = newController.join_phams(toJoin)
if append:
###print 'adding pham', p, 'with members:', p.members
try: newController.add_pham(p)
except: pham.DuplicatePhamError
newController = newController - oldController
print "there are", len(newController.phams), "new phams."
for newPham in newController.phams: print newPham
print "Saving..."
newController.save()
def get_fasta_from_pham(self, phamName): self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor() self.db = pham.db(c = self.c) return phamerator_manage_db.get_fasta_from_pham(self.c, phamName)
def get_relatives(self, GeneID, alignmentType='both', blastThreshold=None, clustalwThreshold=None): self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor() self.db = pham.db(c = self.c) return phamerator_manage_db.get_relatives(self.c, GeneID, alignmentType=alignmentType, blastThreshold=blastThreshold, clustalwThreshold=clustalwThreshold)
def get_scores(self, query, subject): self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor() self.db = pham.db(c = self.c) return phamerator_manage_db.get_scores(self.c, query, subject)
def get_gene_start_stop_length_orientation_from_GeneID(self, GeneID): self.c = db_conf.db_conf(username=self.username,password=self.password,server=self.server,db=self.database).get_cursor() self.db = pham.db(c = self.c) return phamerator_manage_db.get_gene_start_stop_length_orientation_from_GeneID(self.c, GeneID)
def create_pham_circle(self, phamName, alignmentColor, adjustment, radius):
self.c = db_conf.db_conf(username=self.username,
password=self.password,
server=self.server,
db=self.database).get_cursor()
self.db = pham.db(c=self.c)
filename = str(phamName) + str(alignmentColor) + str(adjustment) + str(
radius)
filelist = os.listdir(self.cachedir)
for item in filelist:
if item == filename:
self.current_phamCircle = self.cachedir + filename
string = "phamCircle " + str(
phamName) + " found in cache, using cached version\n"
print ">" + string
return string
self.radius = radius
self.alignmentColor = alignmentColor
self.threshold = 0.70
self.phamCircleCanvas = goocanvas.Canvas()
GeneIDs = phamerator_manage_db.get_members_of_pham(self.c, phamName)
memberPhages, self.nonMemberPhages = [], []
for GeneID in GeneIDs:
PhageID = phamerator_manage_db.get_PhageID_from_GeneID(
self.c, GeneID)
if PhageID not in memberPhages: memberPhages.append(PhageID)
totalPhages = phamerator_manage_db.get_PhageIDs(self.c)
for p in totalPhages:
if p not in memberPhages:
self.nonMemberPhages.append(
phamerator_manage_db.get_phage_name_from_PhageID(
self.c, p))
self.l = []
self.genes = []
for a in GeneIDs:
for b in GeneIDs:
if a != b:
for gene in [a, b]:
if gene not in self.genes: self.genes.append(gene)
clustalwScore, blastScore = phamerator_manage_db.get_scores(
self.c, a, b)
if clustalwScore >= 0.275:
self.l.append((a, b, 'clustalw', clustalwScore))
if blastScore and blastScore <= 0.0001:
self.l.append((a, b, 'blast', blastScore))
self.phamCircle = PhamDisplay.PhamCircle(phamName,
self.c,
verbose=True,
radius=self.radius)
if self.alignmentColor == True:
self.phamCircleCanvas.set_root_item_model(
self.phamCircle.create_canvas_model(self.nonMemberPhages,
self.genes,
self.l,
adjustment,
self.threshold,
blastColor='#ff0000',
clustalwColor='#0000ff'))
else:
phamColorFromDataBase = self.db.select('pham_color',
'color',
name=phamName)[0][0]
self.phamCircleCanvas.set_root_item_model(
self.phamCircle.create_canvas_model(
self.nonMemberPhages,
self.genes,
self.l,
adjustment,
self.threshold,
allColor=phamColorFromDataBase))
"""x, y = (600, 500)
self.phamCircleCanvas.set_size_request(x, y)
self.defaultPhamCircleCanvasSize = (x, y)
self.phamCircleCanvas.show()
self.window.window.set_cursor(None)
return False"""
self.current_phamCircle = self.cachedir + filename
self.phamCircleCanvas.set_bounds(0, 0, 10000, 10000)
surface = cairo.SVGSurface(self.current_phamCircle, 15 * 72, 15 * 72)
cr = cairo.Context(surface)
cr.translate(10, 0)
self.phamCircleCanvas.render(cr, None, 0.1)
cr.show_page()
string = "phamCircle " + str(phamName) + " generated\n"
print ">" + string
return string
def get_pham(self):
db = pham.db(c)
pham = db.select("pham", name, GeneID=self.GeneID)
return pham
