def site(variant_list, tx, ub_intervals, nuc_context=3, num_simulations=10000):
# interpet variant context
var_sub, dna_change_sub, trinuc_context = sc.get_substitution_trinuc_context(
variant_list, tx)
#trinuc_context = [sc.get_chasm_context(nc) for nc in trinuc_context]
trinuc_count = collections.Counter(
trinuc_context).items() # count the trinucleotides
# return if no substitution
if not var_sub:
return None
# create sequence context obj
seq_context = sc.SequenceContext(tx, nuc_context)
# figure out how many overlap sites of interest
num_ub_mut = 0
for myvariant in var_sub:
num_ub_mut += utils.overlap_with_intervals(myvariant, ub_intervals)
# return p-value of 1 if no degrons affected
if num_ub_mut == 0:
return 0, 1
# simulate new mutations
random_sub = seq_context.random_pos(trinuc_count, num_simulations)
# evaluate the simulations
ub_ct, iter_sim = 0, 0
tmp_mut_pos = np.hstack([abs_pos for ctxt, cds_pos, abs_pos in random_sub])
for sim_pos in tmp_mut_pos:
# get the variant effect for simulated mutations
sim_variant_subs = variants.get_mutation_info(sim_pos, tx,
dna_change_sub)
# count the degron mutations
sim_num_mut = 0
for sim_variant in sim_variant_subs:
sim_num_mut += utils.overlap_with_intervals(
sim_variant, ub_intervals)
# count if exceeds observed statistic
if sim_num_mut >= num_ub_mut:
ub_ct += 1
# update iteration counter
iter_sim += 1
# stop if sufficient number of simulations reached
if ub_ct >= 100:
break
# calculate p-value
ub_pval = ub_ct / float(iter_sim)
return num_ub_mut, ub_pval
def score_raw(variant_list, tx, clf1, clf2, model='cterm', nuc_context=1.5):
"""Simulate the effect of mutations on terminal degrons. Handles both c-terminal and and n-terminal degrons.
"""
# interpet variant context
var_sub, dna_change_sub, trinuc_context = sc.get_substitution_trinuc_context(
variant_list, tx)
trinuc_context = [sc.get_chasm_context(nc) for nc in trinuc_context]
trinuc_count = collections.Counter(
trinuc_context).items() # count the trinucleotides
var_sub_no_nmd = utils.filter_nmd_subs(var_sub, tx)
# filter for indel variants
var_indel = [
x for x in variant_list
if x.__class__ in utils.indels and x.variant.is_indel
]
dna_change_indel = [[
v.variant.contig, v.variant.start, v.variant.ref, v.variant.alt
] for v in var_indel]
# skip dropping based on nmd
#var_indel = utils.nmd(var_indel, tx, drop=True)
# return if no substitution
if not var_sub and not var_indel:
return []
# figure out the affect on cterminal degrons
myvars, delta_prob, prob = degron_pred.delta_prob_raw(var_sub_no_nmd +
var_indel,
tx,
clf1,
clf2,
model=model,
is_sum=False)
# skip if no terminal variants
if not len(delta_prob):
return []
# fill in info about each variant
var_info = []
for ix, var in enumerate(myvars):
tmp = [
var.gene_name, var.transcript_id, var.variant.contig,
var.variant.start, var.variant.end, var.short_description,
delta_prob.iloc[ix], prob.iloc[ix]
]
var_info.append(tmp)
return var_info
def degron(variant_list,
tx,
deg_intervals,
nuc_context=3,
num_simulations=10000):
# interpet variant context
var_sub, dna_change_sub, trinuc_context = sc.get_substitution_trinuc_context(
variant_list, tx)
#trinuc_context = [sc.get_chasm_context(nc) for nc in trinuc_context]
trinuc_count = collections.Counter(
trinuc_context).items() # count the trinucleotides
# return if no substitution
if not var_sub:
return None
# create sequence context obj
seq_context = sc.SequenceContext(tx, nuc_context)
# figure out how many affect lysines
num_deg_mut = 0
for myvariant in var_sub:
num_deg_mut += utils.overlap_with_intervals(myvariant, deg_intervals)
# return p-value of 1 if no degrons affected
if num_deg_mut == 0:
return 0, 1
# simulate new mutations
random_sub = seq_context.random_pos(trinuc_count, num_simulations)
# evaluate the simulations
deg_ct_list = []
tmp_mut_pos = np.hstack([abs_pos for ctxt, cds_pos, abs_pos in random_sub])
for sim_pos in tmp_mut_pos:
# get the variant effect for simulated mutations
sim_variant_subs = variants.get_mutation_info(sim_pos, tx,
dna_change_sub)
# count the degron mutations
sim_num_mut = 0
for sim_variant in sim_variant_subs:
sim_num_mut += utils.overlap_with_intervals(
sim_variant, deg_intervals)
deg_ct_list.append(sim_num_mut)
# calculate p-value
deg_pval = np.sum(
1 for x in deg_ct_list if x >= num_deg_mut) / float(num_simulations)
return num_deg_mut, deg_pval
def degron_with_indel(variant_list,
tx,
deg_intervals,
nuc_context=3,
num_simulations=10000):
# interpet variant context
var_sub, dna_change_sub, trinuc_context = sc.get_substitution_trinuc_context(
variant_list, tx)
#trinuc_context = [sc.get_chasm_context(nc) for nc in trinuc_context]
trinuc_count = collections.Counter(
trinuc_context).items() # count the trinucleotides
var_sub_no_nmd = utils.filter_nmd_subs(var_sub, tx)
# filter for indel variants
var_indel = [
x for x in variant_list
if x.__class__ in utils.indels and x.variant.is_indel
]
dna_change_indel = [[
v.variant.contig, v.variant.start, v.variant.ref, v.variant.alt
] for v in var_indel]
var_indel = utils.nmd(var_indel, tx, drop=True)
# return if no substitution
if not var_sub and not var_indel:
return None
# create sequence context obj
seq_context = sc.SequenceContext(tx, nuc_context)
seq_context_indel = sc.SequenceContext(tx, 0)
# figure out how many affect lysines
num_deg_mut = 0
for myvariant in var_sub_no_nmd + var_indel:
num_deg_mut += utils.overlap_with_intervals(myvariant, deg_intervals)
# return p-value of 1 if no degrons affected
if num_deg_mut == 0:
return 0, 1
# simulate new mutations
if var_sub:
random_sub = seq_context.random_pos(trinuc_count, num_simulations)
tmp_mut_pos = np.hstack(
[abs_pos for ctxt, cds_pos, abs_pos in random_sub])
if var_indel:
tmp_input = [('None', len(dna_change_indel))]
random_indel = seq_context_indel.random_pos(tmp_input, num_simulations)
tmp_mut_pos_indel = random_indel[0][2]
# evaluate the simulations
num_sub, num_indel = len(dna_change_sub), len(dna_change_indel)
degron_ct, iter_sim = 0, 0
#for sim_pos in tmp_mut_pos:
for i in range(num_simulations):
# get the variant effect for simulated mutations
if var_sub_no_nmd:
sim_pos = tmp_mut_pos[i, :]
sim_variant_subs = variants.get_mutation_info(
sim_pos, tx, dna_change_sub)
num_sim_subs = len(sim_variant_subs)
sim_variant_subs = utils.filter_nmd_subs(sim_variant_subs, tx)
else:
num_sim_subs = 0
sim_variant_subs = []
# get infor for indel
if var_indel:
sim_pos_indel = tmp_mut_pos_indel[i, :]
sim_variant_indel = variants.get_mutation_info(
sim_pos_indel, tx, dna_change_indel)
num_sim_indels = len(sim_variant_indel)
sim_variant_indel = utils.nmd(sim_variant_indel, tx, drop=True)
else:
num_sim_indels = 0
sim_variant_indel = []
# combine together simulated variants
sim_combined = sim_variant_subs + sim_variant_indel
if not sim_combined:
degron_ct += 1
else:
# count the degron mutations
sim_num_mut = 0
for sim_var in sim_combined:
sim_num_mut += utils.overlap_with_intervals(
sim_var, deg_intervals)
if sim_num_mut >= num_deg_mut:
degron_ct += 1
# update number of iterations
iter_sim += 1
# stop if sufficient number of simulations reached
if degron_ct >= 100:
break
# compute p-value
deg_pval = degron_ct / float(iter_sim)
return num_deg_mut, deg_pval
def terminal_degron(variant_list,
tx,
clf1,
clf2,
model='cterm',
nuc_context=1.5,
num_simulations=10000):
"""Simulate the effect of mutations on terminal degrons. Handles both c-terminal and and n-terminal degrons.
"""
# interpet variant context
var_sub, dna_change_sub, trinuc_context = sc.get_substitution_trinuc_context(
variant_list, tx)
trinuc_context = [sc.get_chasm_context(nc) for nc in trinuc_context]
trinuc_count = collections.Counter(
trinuc_context).items() # count the trinucleotides
var_sub_no_nmd = utils.filter_nmd_subs(var_sub, tx)
# filter for indel variants
var_indel = [
x for x in variant_list
if x.__class__ in utils.indels and x.variant.is_indel
]
dna_change_indel = [[
v.variant.contig, v.variant.start, v.variant.ref, v.variant.alt
] for v in var_indel]
var_indel = utils.nmd(var_indel, tx, drop=True)
# return if no substitution
if not var_sub and not var_indel:
return None
# figure out the affect on cterminal degrons
delta_prob = degron_pred.delta_prob(var_sub_no_nmd + var_indel,
tx,
clf1,
clf2,
model=model)
seqs, _ = utils.process_var_seq(var_sub_no_nmd + var_indel, model=model)
num_impactful_muts = len(seqs)
# skip if no terminal variants
if not delta_prob:
return 0, 1, 0
# create sequence context obj
seq_context = sc.SequenceContext(tx, nuc_context)
seq_context_indel = sc.SequenceContext(tx, 0)
# simulate new mutations
if var_sub:
random_sub = seq_context.random_pos(trinuc_count, num_simulations)
tmp_mut_pos = np.hstack(
[abs_pos for ctxt, cds_pos, abs_pos in random_sub])
# there is no sequence context for indels
if var_indel:
#tmp_input = [('None', len(var_indel))]
tmp_input = [('None', len(dna_change_indel))]
random_indel = seq_context_indel.random_pos(tmp_input, num_simulations)
tmp_mut_pos_indel = random_indel[0][2]
# evaluate the simulations
num_sub, num_indel = len(dna_change_sub), len(dna_change_indel)
delta_prob_ct, iter_sim = 0, 0
for i in range(num_simulations):
# get info for substitutions
if var_sub_no_nmd:
sim_pos = tmp_mut_pos[i, :]
sim_variant_subs = variants.get_mutation_info(
sim_pos, tx, dna_change_sub)
num_sim_subs = len(sim_variant_subs)
# filter based on nmd
sim_variant_subs = utils.filter_nmd_subs(sim_variant_subs, tx)
else:
num_sim_subs = 0
sim_variant_subs = []
# get info for indel
if var_indel:
sim_pos_indel = tmp_mut_pos_indel[i, :]
sim_variant_indel = variants.get_mutation_info(
sim_pos_indel, tx, dna_change_indel)
num_sim_indels = len(sim_variant_indel)
# filter based on nmd
sim_variant_indel = utils.nmd(sim_variant_indel, tx, drop=True)
else:
num_sim_indels = 0
sim_variant_indel = []
# combine together the simulated variants
sim_combined = sim_variant_subs + sim_variant_indel
if not sim_combined:
delta_prob_ct += 1
else:
# get scores from simulations
sim_delta_prob = degron_pred.delta_prob(sim_combined,
tx,
clf1,
clf2,
model=model)
# count if exceeds observed statistic
adjust_factor = (num_sub + num_indel) / (num_sim_subs +
num_sim_indels)
#if abs(sim_delta_prob*adjust_factor) >= abs(delta_prob):
if (sim_delta_prob * adjust_factor) <= delta_prob:
delta_prob_ct += 1
# update number of iterations
iter_sim += 1
# stop if sufficient number of simulations reached
if delta_prob_ct >= 100:
break
# compute p-value
delta_prob_pval = delta_prob_ct / float(iter_sim)
return delta_prob, delta_prob_pval, num_impactful_muts