def get_buried_residues(structure, cutoff, tmp_dir, dssp_path):
"""Finds buried residues by using relative solvent accessible surface area.
"""
# get structure id
structure_id = structure.id
all_letters = set(string.ascii_uppercase) | set(string.ascii_lowercase)
# flatten models into a single model due to limitations of DSSP
id_map = {}
for k, model in enumerate(structure):
if k == 0:
#used_letters = set(model.child_dict.keys())
used_letters = set()
for chain in model:
if chain.get_id() == ' ':
chain.id = 'A'
#for l in used_letters:
id_map[(model.id, chain.id)] = (model.id, chain.id)
used_letters.add(chain.id)
new_model = model.id
else:
for chain in model:
left_over = all_letters - used_letters
if not left_over:
# if run out of chain letters just return nothing
return []
new_letter = left_over.pop()
used_letters.add(new_letter)
old_letter = chain.id
chain.id = new_letter
id_map[(new_model, new_letter)] = (model.id, old_letter)
# add numbers if there is not more letters left
if not (all_letters - used_letters):
all_letters.update(
set(string.digits) | set(string.punctuation))
model.id = new_model
# save new structure to tmp dir
io = Bio.PDB.PDBIO()
io.set_structure(structure)
tmp_path = os.path.join(tmp_dir, structure_id + '.pdb')
io.save(tmp_path)
# read in tmp structure
tmp_structure = utils.read_structure(tmp_path, structure_id, quiet=True)
# find the solvent accessibility for residues
dssp_results = Bio.PDB.DSSP(tmp_structure[0], tmp_path, dssp=dssp_path)
# get bfactors for each amino acid residue
bfacs_missing = [
r for r in tmp_structure.get_residues()
if Bio.PDB.is_aa(r) and 'CA' not in r.child_dict
]
bfacs = [
r['CA'].get_bfactor() for r in tmp_structure.get_residues()
if Bio.PDB.is_aa(r) and 'CA' in r.child_dict
]
mean_bfac = np.mean(bfacs)
std_bfac = np.std(bfacs)
# format output
output = []
for result in dssp_results:
# skip if not an amino acid
if not Bio.PDB.is_aa(result[0]):
continue
# format the ID
full_id = result[0].get_full_id()
#if full_id[2] == ' ':
#full_id[2] = 'A'
try:
orig_model_chain = list(id_map[full_id[1:3]])
except:
print full_id, id_map
raise
# fix missing letter for homology models
if orig_model_chain[1] == ' ':
orig_model_chain[1] = 'A'
# record whether it was buried
if 'CA' in result[0].child_dict:
norm_bfactor = (result[0]['CA'].get_bfactor() -
mean_bfac) / std_bfac
else:
norm_bfactor = None
line = [structure_id] + orig_model_chain + [
result[0].id[1], result[3], norm_bfactor
]
if result[3] <= cutoff:
line.append(1)
else:
line.append(0)
output.append(line)
# delete tmp file
if os.path.exists(tmp_path): os.remove(tmp_path)
return output
def summarize_residues(mutations, pdb_info, radius,
rASA, dssp, tmp_dir,
quiet=True):
# iterate over each structure
logger.info('Running of PDB structures . . .')
output = [['structure', 'tumor type', '# buried residues',
'# protein interface residues', '# nucleic acid interface residues',
'total residues', '# buried mutations', '# protien interface mutations',
'# nucleic acid interface mutations', 'total # mutations',
'burial p-value', 'protein interface p-value',
'nucleic acid interface p-value']]
for structure_id in pdb_info:
#if structure_id.startswith('ENSP') or structure_id.startswith('NP_'):
#continue
#print structure_id
# get pdb info
struct_info = pdb_info[structure_id]
pdb_path = struct_info.pop('path')
# read in structure
structure = utils.read_structure(pdb_path, structure_id, quiet=quiet)
if structure is None:
continue
# make a list of all chain letters in structure
struct_chains = []
for k in struct_info.keys():
struct_chains.extend(struct_info[k])
structure_mutations = mutations.get(structure_id, [])
# skip structure if no mutations
if not structure_mutations:
continue
# separate out mutation info
ttypes, mres, mcount, mchains = zip(*structure_mutations) # if model_mutations else ([], [], [])
# stratify mutations by their tumor type
# ttype_ixs is a dictionary that contains
# ttype as the keys and a list of relevant
# indices as the values
unique_ttypes = set(ttypes)
ttype_ixs = {t: [i for i in range(len(mcount)) if ttypes[i]==t]
for t in unique_ttypes}
#ttype_ixs['PANCAN'] = range(len(mcount))
# add PANCAN as a "tumour type"
unique_ttypes = list(unique_ttypes)
#unique_ttypes.append('PANCAN')
# obtain relevant info from structure
tmp_info = pstruct.get_structure_info(structure, mchains, mres, mcount,
struct_chains, ttype_ixs)
(mut_res_centers_of_geometry,
mut_res_mutation_counts,
all_res_centers_of_geometry,
models) = tmp_info
annotated_chains = {chain
for description in struct_info
for chain in struct_info[description]}
# find buried residues
buried_res = pstruct.get_buried_residues(structure, rASA, tmp_dir, dssp)
tmp_buried = [res_id
for res_id in buried_res
if res_id[2] in annotated_chains]
total_res = len(tmp_buried)
buried_res_info = {(info[1], info[2], info[3])
for info in tmp_buried
if info[-1] == 1}
num_buried_res = len(buried_res_info)
# find interface residues for proteins and nucleic acids
interface_res = pstruct.get_interface_residues(structure, radius)
interface_prot_info = {(res_id[1], res_id[2], res_id[3][1])
for res_id in interface_res
if (res_id[2] in annotated_chains) and interface_res[res_id][0]==1}
interface_na_info = {(res_id[1], res_id[2], res_id[3][1])
for res_id in interface_res
if (res_id[2] in annotated_chains) and sum(interface_res[res_id][1:])>=1}
num_interface_prot_res = len(interface_prot_info)
num_interface_na_res = len(interface_na_info)
# iterate through each tumour type
pan_counts = []
pan_buried_counts = []
pan_interface_prot_counts, pan_interface_na_counts = [], []
tmp_output = []
for tumour in unique_ttypes:
# skip tumor types if not one specified
#if (not opts['tumor_type'] == tumour and not opts['tumor_type'] == 'EVERY'):
#continue
# draw information for the specific tumour type
t_mut_res_centers_of_geometry = mut_res_centers_of_geometry[tumour]
t_mut_res_mutation_counts = mut_res_mutation_counts[tumour]
# count total mutations in structure while
# avoiding double counting due to same id and chain
# being on multiple models
obs_models = []
obs_chains = []
total_mutations = 0
total_buried_muts = 0
total_interface_prot_muts, total_interface_na_muts = 0, 0
banned_chains = set()
#if not tumour == 'PANCAN':
if True:
for k in t_mut_res_mutation_counts:
mutations_to_add = t_mut_res_mutation_counts[k]
# prevent double counting
cur_model = k[1]
cur_chain = k[2]
cur_pos = k[3][1]
for i in range(len(obs_models)):
if not cur_model == obs_models[i] and cur_chain == obs_chains[i]:
mutations_to_add = 0
break
if (cur_chain, cur_pos) in banned_chains:
mutations_to_add = 0
# add all equivalent chains to banned list
equiv_chains = pstruct.find_eq_letters(struct_info, cur_chain)
if equiv_chains is not None:
equiv_pos = set([(e, cur_pos) for e in equiv_chains])
banned_chains |= equiv_pos - set([(cur_chain, cur_pos)])
# add to total mutation count
total_mutations += mutations_to_add
# current residue of interest
curr_res = (cur_model, cur_chain, cur_pos)
# add buried residue mutation counts
is_buried = [(m, c[0], c[1]) in buried_res_info
for c in equiv_pos
for m in range(4)]
#if (curr_res in buried_res_info):
if any(is_buried):
total_buried_muts += mutations_to_add
pan_buried_counts.append(mutations_to_add)
# add interface residue mutation counts
is_interface_prot = [(m, c[0], c[1]) in interface_prot_info
for c in equiv_pos
for m in range(4)]
is_interface_na = [(m, c[0], c[1]) in interface_na_info
for c in equiv_pos
for m in range(4)]
#if (curr_res in interface_info):
if any(is_interface_prot):
total_interface_prot_muts += mutations_to_add
pan_interface_prot_counts.append(mutations_to_add)
if any(is_interface_na):
total_interface_na_muts += mutations_to_add
pan_interface_na_counts.append(mutations_to_add)
# mark chains/models
obs_models.append(k[1])
obs_chains.append(k[2])
pan_counts.append(total_mutations)
else:
total_mutations = sum(pan_counts)
total_buried_muts = sum(pan_buried_counts)
total_interface_prot_muts = sum(pan_interface_prot_counts)
total_interface_na_muts = sum(pan_interface_na_counts)
tmp_output.append([structure_id, tumour, num_buried_res,
num_interface_prot_res, num_interface_na_res,
total_res, total_buried_muts,
total_interface_prot_muts, total_interface_na_muts,
total_mutations,
])
output.extend(tmp_output)
return output
def get_buried_residues(structure, cutoff, tmp_dir, dssp_path):
"""Finds buried residues by using relative solvent accessible surface area.
"""
# get structure id
structure_id = structure.id
all_letters = set(string.ascii_uppercase) | set(string.ascii_lowercase)
# flatten models into a single model due to limitations of DSSP
id_map = {}
for k, model in enumerate(structure):
if k == 0:
#used_letters = set(model.child_dict.keys())
used_letters = set()
for chain in model:
if chain.get_id() == ' ':
chain.id = 'A'
#for l in used_letters:
id_map[(model.id, chain.id)] = (model.id, chain.id)
used_letters.add(chain.id)
new_model = model.id
else:
for chain in model:
left_over = all_letters - used_letters
if not left_over:
# if run out of chain letters just return nothing
return []
new_letter = left_over.pop()
used_letters.add(new_letter)
old_letter = chain.id
chain.id = new_letter
id_map[(new_model, new_letter)] = (model.id, old_letter)
# add numbers if there is not more letters left
if not (all_letters - used_letters):
all_letters.update(set(string.digits) | set(string.punctuation))
model.id = new_model
# save new structure to tmp dir
io = Bio.PDB.PDBIO()
io.set_structure(structure)
tmp_path = os.path.join(tmp_dir, structure_id+'.pdb')
io.save(tmp_path)
# read in tmp structure
tmp_structure = utils.read_structure(tmp_path, structure_id, quiet=True)
# find the solvent accessibility for residues
dssp_results = Bio.PDB.DSSP(tmp_structure[0], tmp_path, dssp=dssp_path)
# get bfactors for each amino acid residue
bfacs_missing = [r
for r in tmp_structure.get_residues()
if Bio.PDB.is_aa(r) and 'CA' not in r.child_dict]
bfacs = [r['CA'].get_bfactor()
for r in tmp_structure.get_residues()
if Bio.PDB.is_aa(r) and 'CA' in r.child_dict]
mean_bfac = np.mean(bfacs)
std_bfac = np.std(bfacs)
# format output
output = []
for result in dssp_results:
# skip if not an amino acid
if not Bio.PDB.is_aa(result[0]):
continue
# format the ID
full_id = result[0].get_full_id()
#if full_id[2] == ' ':
#full_id[2] = 'A'
try:
orig_model_chain = list(id_map[full_id[1:3]])
except:
print full_id, id_map
raise
# fix missing letter for homology models
if orig_model_chain[1] == ' ':
orig_model_chain[1] = 'A'
# record whether it was buried
if 'CA' in result[0].child_dict:
norm_bfactor = (result[0]['CA'].get_bfactor() - mean_bfac) / std_bfac
else:
norm_bfactor = None
line = [structure_id] + orig_model_chain + [result[0].id[1], result[3], norm_bfactor]
if result[3] <= cutoff:
line.append(1)
else:
line.append(0)
output.append(line)
# delete tmp file
if os.path.exists(tmp_path): os.remove(tmp_path)
return output
