def cbgs_identical_pacbp_analysis(first, second):
"""
Obtain data about identity between PacbPs in 2 CodingBlockGraphs
@type first: CodingBlockGraph
@param first: CodingBlockGraph object
@type second: CodingBlockGraph
@param second: CodingBlockGraph object
@rtype: {}, {}
@return: nodes_seen, max_nodes_seen
@attention: nodes_seen counts how often nodes' PacbPs are identical
@attention: max_nodes_seen counts how often nodes' edges exist in both CBGs
"""
# gather list of nodes present in both CBGs
mutual_nodes = comparison.mutual_nodes(first, second)
# set counter dicts for:
# nodes_seen -> nodes with identical PacbPs in both CBGs
# max_node_seen -> nodes that have edges between them in both CBGs
nodes_seen = {}
max_node_seen = {}
for node in mutual_nodes:
nodes_seen[node] = 0
max_node_seen[node] = 0
# loop over all node combinations / edges
for nodeQ, nodeS in first.pairwisecrosscombinations_node():
# only deal with combinations of mutual nodes
if nodeQ not in mutual_nodes or nodeS not in mutual_nodes:
continue
# get PacbP(ORF) of first (donor) and second (acceptor) CBG
donorPacbp, accepPacbp = None, None
if first.has_edge(nodeQ, nodeS):
donorPacbp = first.get_pacbps_by_nodes(node1=nodeQ, node2=nodeS)[0]
if second.has_edge(nodeQ, nodeS):
accepPacbp = second.get_pacbps_by_nodes(node1=nodeQ,
node2=nodeS)[0]
# check if both edges exist -> increase max_node_seen
if accepPacbp and donorPacbp:
max_node_seen[nodeQ] += 1
max_node_seen[nodeS] += 1
# check if Pacbps are identical -> increase nodes_seen
if donorPacbp.barcode() == accepPacbp.barcode():
nodes_seen[nodeQ] += 1
nodes_seen[nodeS] += 1
# return both dictionaries
return nodes_seen, max_node_seen
# end of function cbgs_identical_pacbp_analysis
def cbgs_identical_pacbp_analysis(first,second):
"""
Obtain data about identity between PacbPs in 2 CodingBlockGraphs
@type first: CodingBlockGraph
@param first: CodingBlockGraph object
@type second: CodingBlockGraph
@param second: CodingBlockGraph object
@rtype: {}, {}
@return: nodes_seen, max_nodes_seen
@attention: nodes_seen counts how often nodes' PacbPs are identical
@attention: max_nodes_seen counts how often nodes' edges exist in both CBGs
"""
# gather list of nodes present in both CBGs
mutual_nodes = comparison.mutual_nodes(first,second)
# set counter dicts for:
# nodes_seen -> nodes with identical PacbPs in both CBGs
# max_node_seen -> nodes that have edges between them in both CBGs
nodes_seen = {}
max_node_seen = {}
for node in mutual_nodes:
nodes_seen[node] = 0
max_node_seen[node] = 0
# loop over all node combinations / edges
for nodeQ,nodeS in first.pairwisecrosscombinations_node():
# only deal with combinations of mutual nodes
if nodeQ not in mutual_nodes or nodeS not in mutual_nodes:
continue
# get PacbP(ORF) of first (donor) and second (acceptor) CBG
donorPacbp, accepPacbp = None, None
if first.has_edge(nodeQ,nodeS):
donorPacbp = first.get_pacbps_by_nodes(node1=nodeQ,node2=nodeS)[0]
if second.has_edge(nodeQ,nodeS):
accepPacbp = second.get_pacbps_by_nodes(node1=nodeQ,node2=nodeS)[0]
# check if both edges exist -> increase max_node_seen
if accepPacbp and donorPacbp:
max_node_seen[nodeQ]+=1
max_node_seen[nodeS]+=1
# check if Pacbps are identical -> increase nodes_seen
if donorPacbp.barcode() == accepPacbp.barcode():
nodes_seen[nodeQ]+=1
nodes_seen[nodeS]+=1
# return both dictionaries
return nodes_seen, max_node_seen
# end of function cbgs_identical_pacbp_analysis
def cbgs_identical_pacbp_analysis(first,second):
"""
CBG PacbPORF comparision
@type first: CodingBlockGraph
@param first: CodingBlockGraph
@type second: CodingBlockGraph
@param second: CodingBlockGraph
@rtype: {}, {}
@return: dicts with counts how often pacbps of mutual nodes are identical
"""
# gather list of nodes present in both CBGs
mutual_node_list = mutual_nodes(first,second)
# set counter dicts for:
# nodes_seen -> nodes with identical PacbPs in both CBGs
# max_node_seen -> nodes that have edges between them in both CBGs
nodes_seen = {}
max_node_seen = {}
mutual_pacbp_edges = []
for node in mutual_node_list:
nodes_seen[node] = 0
max_node_seen[node] = 0
# loop over all node combinations / edges
for nodeQ,nodeS in first.pairwisecrosscombinations_node():
# only deal with combinations of mutual nodes
if nodeQ not in mutual_node_list or nodeS not in mutual_node_list:
continue
# get PacbP(ORF) of first (donor) and second (acceptor) CBG
donorPacbp, accepPacbp = None, None
if first.has_edge(nodeQ,nodeS):
donorPacbp = first.get_pacbps_by_nodes(
node1=nodeQ,node2=nodeS)[0]
if second.has_edge(nodeQ,nodeS):
accepPacbp = second.get_pacbps_by_nodes(
node1=nodeQ,node2=nodeS)[0]
# check if both edges exist -> increase max_node_seen
if accepPacbp and donorPacbp:
max_node_seen[nodeQ]+=1
max_node_seen[nodeS]+=1
# check if Pacbps are identical -> increase nodes_seen
if donorPacbp.barcode() == accepPacbp.barcode():
mutual_pacbp_edges.append( (nodeQ,nodeS) )
nodes_seen[nodeQ]+=1
nodes_seen[nodeS]+=1
# return both dictionaries
return nodes_seen, max_node_seen
def project_missing_introns(cbgs,
interface_intron_status={},
inframe_intron_min_aa_length=INFRAME_INTRON_MIN_AA_LENGTH,
verbose=False):
"""
Create ProjectedIntrons for intron gain/loss events in ordered CBGs
@type cbgs: list
@param cbgs: ordered (!) list of CodingBlockGraphs
@type interface_intron_status: dict
@param interface_intron_status:
@type inframe_intron_min_aa_length: integer
@param inframe_intron_min_aa_length: minimal AA length of an inframe intron
@type verbose: Boolean
@param verbose: print logging/debugging messages to STDOUT when True
@rtype: dict
@return: dict with organism as keys, list of Projected Introns as values
"""
# fill dictionary projected_introns with keys of each org identifier
return_projected_introns= {}
for node in cbgs[0].get_nodes():
return_projected_introns[cbgs[0]._organism_from_node(node)] = []
# loop over all other coding blocks
# and evaluate the junction between `prev` and `this`
for position in range(1,len(cbgs)):
this = cbgs[position]
prev = cbgs[position-1]
# find intersection nodes between graphs
intersectnodes = mutual_nodes(prev,this)
if intersectnodes:
# yep, there are shared ORFs between the adjacent CBGs
# this means we can try to project introns
intersectorgs =\
[ prev._organism_from_node(node) for node in intersectnodes ]
# check for identical pacbps shared by both CBGs ->
# no intron projection here!
nodes_seen,max_nodes_seen = cbgs_identical_pacbp_analysis(prev,this)
####################################################################
if verbose:
print "nodes:", nodes_seen
print "inters", max_nodes_seen, "intersectorgs:", intersectorgs
####################################################################
for orgQ,orgS in prev.pairwisecrosscombinations_organism():
# check interface_intron_status dict. Do only a projection
# when orgQ status in (True,None) and orgS status in
# (False,None) or vice versa
if interface_intron_status:
if interface_intron_status[orgQ] in (True,None) and\
interface_intron_status[orgS] in (False,None):
pass
elif interface_intron_status[orgS] in (True,None) and\
interface_intron_status[orgQ] in (False,None):
pass
else:
# no aparantly projection required in this interface!
continue
else:
# Interface_intron_status not applied or empty.
# Just pass here, will be solved lateron
pass
# ignore by continue if orgQ or orgS identifier is absent
# in this or prev CBG; this happens when it is a K(s-x) CBG
if orgQ not in prev.organism_set(): continue
if orgQ not in this.organism_set(): continue
if orgS not in prev.organism_set(): continue
if orgS not in this.organism_set(): continue
# get the corresponding nodes
nodeQprev = prev.node_by_organism(orgQ)
nodeSprev = prev.node_by_organism(orgS)
nodeQthis = this.node_by_organism(orgQ)
nodeSthis = this.node_by_organism(orgS)
# check if edge exists in both CBGs (K(s) CBG, no edges missing)
if not prev.has_edge(nodeQprev,nodeSprev) or\
not this.has_edge(nodeQthis,nodeSthis):
continue
# if both organisms are not part of the intersection ->
# 2x non-identical Orfs so, no projection possible here
if not orgQ in intersectorgs and not orgS in intersectorgs:
continue
# calculate distance between codingblocks
distQ = prev.distance_between_codingblocks(this,organism=orgQ)
distS = prev.distance_between_codingblocks(this,organism=orgS)
# check for potential inframe intron; if found,
# ignore the intron projection here
if orgQ in intersectorgs and orgS in intersectorgs:
if nodes_seen[nodeQprev] >= max_nodes_seen[nodeQprev]-1 and\
nodes_seen[nodeSprev] >= max_nodes_seen[nodeSprev]-1:
# The Pacbp of this edge is identical in `prev`
# and `this` CBG. Now check the distance beteen the
# CBGs (distance between their OMSRs). If this distance
# is large or distinct between the 2 organisms/genes,
# then intron projection is still required/usefull.
# If small, then possibly an inframe intron or a region
# of low AA similarity of variable length range.
if abs(distQ-distS) < inframe_intron_min_aa_length:
# distance discrepancy is fairly small; this
# represents a continious alignment.
# Do not project introns here!
continue
# get Pacbp of donor/prev CBG of this organism combination
pfD = prev.get_pacbps_by_nodes(
node1=nodeQprev,node2=nodeSprev)[0]
pfA = this.get_pacbps_by_nodes(
node1=nodeQthis,node2=nodeSthis)[0]
# Now map Projected Introns!
if nodeQprev == nodeQthis and nodeSprev != nodeSthis:
# continious query alignment
tmp_projected_introns = PrjIntLqry(pfD,pfA)
org_of_projection = orgQ
elif nodeSprev == nodeSthis and nodeQprev != nodeQthis:
# continious sbjct alignment
tmp_projected_introns = PrjIntLsbj(pfD,pfA)
org_of_projection = orgS
elif nodeQprev != nodeQthis and nodeSprev != nodeSthis:
# this case is recognized a few lines above:
# not orgQ in intersectorgs and not orgS in intersectorgs
raise "THIS CANNOT HAPPEN!!!"
else:
# intersection nodes/organisms. Several options possible
if distS > distQ and nodes_seen[nodeQprev] >=\
max_nodes_seen[nodeQprev]-1:
# continious query alignment
tmp_projected_introns = PrjIntLqry(pfD,pfA)
org_of_projection = orgQ
elif distQ > distS and nodes_seen[nodeSprev] >=\
max_nodes_seen[nodeSprev]-1:
# continious sbjct alignment
tmp_projected_introns = PrjIntLsbj(pfD,pfA)
org_of_projection = orgS
else:
########################################################
if verbose:
# hmm.... seems an inframe introns in two species!?
print "WARNING: inframe intron 2 species!!!",
print orgQ,orgS, distQ, distS
print nodeQprev, nodeQthis,
print "seen:", nodes_seen[nodeQprev],
print "max:", max_nodes_seen[nodeQthis],
print nodeSprev, nodeSthis,
print "seen:", nodes_seen[nodeSprev],
print "max:", max_nodes_seen[nodeSthis]
########################################################
# no mapping of intron possible because
# no continious query or sbjct
continue
################################################################
if verbose:
print orgQ,orgS, "intron count:", len(tmp_projected_introns)
################################################################
# done! now remap/process the `tmp_projected_introns` and
# store to return_projected_introns dictionary.
# See if it is a new projected intron, or an intron that
# coinsides with a previous found one. If it coinsides, it
# is even stringer proof that this is a valid Projected Intron
# and as a consequence its score will be increased
for proj_intron in tmp_projected_introns:
current_projected_intron = None
# see if there is in current species already an exact
# projected intron of this kind; if so,
# add_projection_intron() to the earlier found intron.
# This will increase the score of this particular intron
for known_projected_intron in\
return_projected_introns[org_of_projection]:
if proj_intron.barcode() ==\
known_projected_intron.barcode():
known_projected_intron.add_projected_intron(
proj_intron.projected_introns[0] )
current_projected_intron = known_projected_intron
break
else:
# not seen before; add to return_projected_introns
return_projected_introns[org_of_projection].append(
proj_intron )
current_projected_intron = proj_intron
# Not projected onto the earlier found introns yet,
# so do it here by add_projected_intron() to them too.
true_intron = proj_intron.projected_introns[0]
for _org in return_projected_introns.keys():
if _org == org_of_projection: continue
for known_projected_intron in\
return_projected_introns[_org]:
if true_intron.barcode() in\
[ i.barcode() for i in\
known_projected_intron.projected_introns ]:
# append this intron as extra evidence in
# this other organism
known_projected_intron.add_projected_intron(
true_intron )
# append this same intron another time
# to this organism
current_projected_intron.add_projected_intron(
true_intron )
########################################################################
if verbose:
print "SUMMED INTRONS PER ORG:", (position-1,position),
print [ (org,len(return_projected_introns[org])) for org\
in return_projected_introns.keys() ]
print interface_intron_status
########################################################################
# sort the projected introns by total_score
for org in return_projected_introns.keys():
tmp = []
for projintron in return_projected_introns[org]:
projD = projintron.get_projected_donor_site()
projA = projintron.get_projected_acceptor_site()
tmp.append( ( projintron.total_score(), projintron ) )
tmp.sort()
tmp.reverse()
return_projected_introns[org] = [ proj for score,proj in tmp ]
# and return the dictionary of projected_introns
return return_projected_introns
def project_missing_introns(cbgs,
interface_intron_status={},
inframe_intron_min_aa_length=INFRAME_INTRON_MIN_AA_LENGTH,
verbose=False):
"""
Create ProjectedIntrons for intron gain/loss events in ordered CBGs
@type cbgs: list
@param cbgs: ordered (!) list of CodingBlockGraphs
@type interface_intron_status: dict
@param interface_intron_status:
@type inframe_intron_min_aa_length: integer
@param inframe_intron_min_aa_length: minimal AA length of an inframe intron
@type verbose: Boolean
@param verbose: print logging/debugging messages to STDOUT when True
@rtype: dict
@return: dict with organism as keys, list of Projected Introns as values
"""
# fill dictionary projected_introns with keys of each org identifier
return_projected_introns= {}
for node in cbgs[0].get_nodes():
return_projected_introns[cbgs[0]._organism_from_node(node)] = []
# loop over all other coding blocks
# and evaluate the junction between `prev` and `this`
for position in range(1,len(cbgs)):
this = cbgs[position]
prev = cbgs[position-1]
# find intersection nodes between graphs
intersectnodes = mutual_nodes(prev,this)
if intersectnodes:
# yep, there are shared ORFs between the adjacent CBGs
# this means we can try to project introns
intersectorgs =\
[ prev._organism_from_node(node) for node in intersectnodes ]
# check for identical pacbps shared by both CBGs ->
# no intron projection here!
nodes_seen,max_nodes_seen = cbgs_identical_pacbp_analysis(prev,this)
####################################################################
if verbose:
print "nodes:", nodes_seen
print "inters", max_nodes_seen, "intersectorgs:", intersectorgs
####################################################################
for orgQ,orgS in prev.pairwisecrosscombinations_organism():
# check interface_intron_status dict. Do only a projection
# when orgQ status in (True,None) and orgS status in
# (False,None) or vice versa
if interface_intron_status:
if interface_intron_status[orgQ] in (True,None) and\
interface_intron_status[orgS] in (False,None):
pass
elif interface_intron_status[orgS] in (True,None) and\
interface_intron_status[orgQ] in (False,None):
pass
else:
# no aparantly projection required in this interface!
continue
else:
# Interface_intron_status not applied or empty.
# Just pass here, will be solved lateron
pass
# ignore by continue if orgQ or orgS identifier is absent
# in this or prev CBG; this happens when it is a K(s-x) CBG
if orgQ not in prev.organism_set(): continue
if orgQ not in this.organism_set(): continue
if orgS not in prev.organism_set(): continue
if orgS not in this.organism_set(): continue
# get the corresponding nodes
nodeQprev = prev.node_by_organism(orgQ)
nodeSprev = prev.node_by_organism(orgS)
nodeQthis = this.node_by_organism(orgQ)
nodeSthis = this.node_by_organism(orgS)
# check if edge exists in both CBGs (K(s) CBG, no edges missing)
if not prev.has_edge(nodeQprev,nodeSprev) or\
not this.has_edge(nodeQthis,nodeSthis):
continue
# if both organisms are not part of the intersection ->
# 2x non-identical Orfs so, no projection possible here
if not orgQ in intersectorgs and not orgS in intersectorgs:
continue
# calculate distance between codingblocks
distQ = prev.distance_between_codingblocks(this,organism=orgQ)
distS = prev.distance_between_codingblocks(this,organism=orgS)
# check for potential inframe intron; if found,
# ignore the intron projection here
if orgQ in intersectorgs and orgS in intersectorgs:
if nodes_seen[nodeQprev] >= max_nodes_seen[nodeQprev]-1 and\
nodes_seen[nodeSprev] >= max_nodes_seen[nodeSprev]-1:
# The Pacbp of this edge is identical in `prev`
# and `this` CBG. Now check the distance beteen the
# CBGs (distance between their OMSRs). If this distance
# is large or distinct between the 2 organisms/genes,
# then intron projection is still required/usefull.
# If small, then possibly an inframe intron or a region
# of low AA similarity of variable length range.
if abs(distQ-distS) < inframe_intron_min_aa_length:
# distance discrepancy is fairly small; this
# represents a continious alignment.
# Do not project introns here!
continue
# get Pacbp of donor/prev CBG of this organism combination
pfD = prev.get_pacbps_by_nodes(
node1=nodeQprev,node2=nodeSprev)[0]
pfA = this.get_pacbps_by_nodes(
node1=nodeQthis,node2=nodeSthis)[0]
# Now map Projected Introns!
if nodeQprev == nodeQthis and nodeSprev != nodeSthis:
# continious query alignment
tmp_projected_introns = PrjIntLqry(pfD,pfA)
org_of_projection = orgQ
elif nodeSprev == nodeSthis and nodeQprev != nodeQthis:
# continious sbjct alignment
tmp_projected_introns = PrjIntLsbj(pfD,pfA)
org_of_projection = orgS
elif nodeQprev != nodeQthis and nodeSprev != nodeSthis:
# this case is recognized a few lines above:
# not orgQ in intersectorgs and not orgS in intersectorgs
raise "THIS CANNOT HAPPEN!!!"
else:
# intersection nodes/organisms. Several options possible
if distS > distQ and nodes_seen[nodeQprev] >=\
max_nodes_seen[nodeQprev]-1:
# continious query alignment
tmp_projected_introns = PrjIntLqry(pfD,pfA)
org_of_projection = orgQ
elif distQ > distS and nodes_seen[nodeSprev] >=\
max_nodes_seen[nodeSprev]-1:
# continious sbjct alignment
tmp_projected_introns = PrjIntLsbj(pfD,pfA)
org_of_projection = orgS
else:
########################################################
if verbose:
# hmm.... seems an inframe introns in two species!?
print "WARNING: inframe intron 2 species!!!",
print orgQ,orgS, distQ, distS
print nodeQprev, nodeQthis,
print "seen:", nodes_seen[nodeQprev],
print "max:", max_nodes_seen[nodeQthis],
print nodeSprev, nodeSthis,
print "seen:", nodes_seen[nodeSprev],
print "max:", max_nodes_seen[nodeSthis]
########################################################
# no mapping of intron possible because
# no continious query or sbjct
continue
################################################################
if verbose:
print orgQ,orgS, "intron count:", len(tmp_projected_introns)
################################################################
# done! now remap/process the `tmp_projected_introns` and
# store to return_projected_introns dictionary.
# See if it is a new projected intron, or an intron that
# coinsides with a previous found one. If it coinsides, it
# is even stringer proof that this is a valid Projected Intron
# and as a consequence its score will be increased
for proj_intron in tmp_projected_introns:
current_projected_intron = None
# see if there is in current species already an exact
# projected intron of this kind; if so,
# add_projection_intron() to the earlier found intron.
# This will increase the score of this particular intron
for known_projected_intron in\
return_projected_introns[org_of_projection]:
if proj_intron.barcode() ==\
known_projected_intron.barcode():
known_projected_intron.add_projected_intron(
proj_intron.projected_introns[0] )
current_projected_intron = known_projected_intron
break
else:
# not seen before; add to return_projected_introns
return_projected_introns[org_of_projection].append(
proj_intron )
current_projected_intron = proj_intron
# Not projected onto the earlier found introns yet,
# so do it here by add_projected_intron() to them too.
true_intron = proj_intron.projected_introns[0]
for _org in return_projected_introns.keys():
if _org == org_of_projection: continue
for known_projected_intron in\
return_projected_introns[_org]:
if true_intron.barcode() in\
[ i.barcode() for i in\
known_projected_intron.projected_introns ]:
# append this intron as extra evidence in
# this other organism
known_projected_intron.add_projected_intron(
true_intron )
# append this same intron another time
# to this organism
current_projected_intron.add_projected_intron(
true_intron )
########################################################################
if verbose:
print "SUMMED INTRONS PER ORG:", (position-1,position),
print [ (org,len(return_projected_introns[org])) for org\
in return_projected_introns.keys() ]
print interface_intron_status
