def fitness_freq(virus):
if virus == 'h1':
from combinatorial_fitness import load_doud2016
strains, seqs_fitness = load_doud2016()
strain = 'h1'
train_fname = 'target/flu/clusters/all_h1.fasta'
mut_fname = 'target/flu/mutation/mutations_h1.fa'
anchor_id = ('gb:LC333185|ncbiId:BBB04702.1|UniProtKB:-N/A-|'
'Organism:Influenza')
elif virus == 'bf520':
from combinatorial_fitness import load_haddox2018
strains, seqs_fitness = load_haddox2018()
strain = 'BF520'
train_fname = 'target/hiv/clusters/all_BF520.fasta'
mut_fname = 'target/hiv/mutation/mutations_bf520.fa'
anchor_id = 'A1.KE.1994.BF520.W14M.C2.KX168094'
elif virus == 'bg505':
from combinatorial_fitness import load_haddox2018
strains, seqs_fitness = load_haddox2018()
strain = 'BG505'
train_fname = 'target/hiv/clusters/all_BG505.fasta'
mut_fname = 'target/hiv/mutation/mutations_hiv.fa'
anchor_id = 'A1.KE.-.BG505_W6M_ENV_C2.DQ208458'
else:
raise ValueError('invalid option {}'.format(virus))
anchor = None
pos_aa_freq = {}
for idx, record in enumerate(SeqIO.parse(train_fname, 'fasta')):
mutation = record.seq
if record.id == anchor_id:
anchor = mutation
else:
for pos, c in enumerate(mutation):
if c == '-':
continue
if (pos, c) not in pos_aa_freq:
pos_aa_freq[(pos, c)] = 0.
pos_aa_freq[(pos, c)] += 1.
assert (anchor is not None)
mutations = [str(record.seq) for record in SeqIO.parse(mut_fname, 'fasta')]
mut_freqs, fitnesses = [], []
for mut_idx, mutation in enumerate(mutations):
if mutation == anchor:
continue
mutation_clean = str(mutation).replace('-', '')
if (mutation_clean, strain) not in seqs_fitness:
continue
didx = [c1 != c2 for c1, c2 in zip(anchor, mutation)].index(True)
if (didx, mutation[didx]) in pos_aa_freq:
mut_freqs.append(pos_aa_freq[(didx, mutation[didx])])
else:
mut_freqs.append(0)
fitnesses.append(seqs_fitness[(mutation_clean,
strain)][0]['preference'])
mut_freqs = np.array(mut_freqs)
print_result(mut_freqs, fitnesses, virus + ' (freq)')
def fitness_energy(virus):
if virus == 'h1':
from combinatorial_fitness import load_doud2016
strains, seqs_fitness = load_doud2016()
strain = 'h1'
train_fname = 'target/flu/clusters/all_h1.fasta'
mut_fname = 'target/flu/mutation/mutations_h1.fa'
energy_fname = 'target/flu/clusters/all_h1.fasta.E.txt'
elif virus == 'bf520':
from combinatorial_fitness import load_haddox2018
strains, seqs_fitness = load_haddox2018()
strain = 'BF520'
train_fname = 'target/hiv/clusters/all_BF520.fasta'
mut_fname = 'target/hiv/mutation/mutations_bf520.fa'
energy_fname = 'target/hiv/clusters/all_BF520.fasta.E.txt'
elif virus == 'bg505':
from combinatorial_fitness import load_haddox2018
strains, seqs_fitness = load_haddox2018()
strain = 'BG505'
train_fname = 'target/hiv/clusters/all_BG505.fasta'
mut_fname = 'target/hiv/mutation/mutations_hiv.fa'
energy_fname = 'target/hiv/clusters/all_BG505.fasta.E.txt'
else:
raise ValueError('invalid option {}'.format(virus))
train_seqs = list(SeqIO.parse(train_fname, 'fasta'))
mut_seqs = list(SeqIO.parse(mut_fname, 'fasta'))
energies = np.loadtxt(energy_fname)
assert (len(energies) == len(train_seqs) + len(mut_seqs))
mut2energy = {
str(seq.seq).replace('-', ''): -energy
for seq, energy in zip(mut_seqs, energies[len(train_seqs):])
}
fitnesses, predictions = [], []
for mut_seq, strain in seqs_fitness:
if mut_seq not in mut2energy:
continue
fitnesses.append(seqs_fitness[(mut_seq, strain)][0]['preference'])
predictions.append(mut2energy[mut_seq])
print_result(fitnesses, predictions, virus + ' (Potts)')
seq_to_mutate, escape_seqs = load_dingens2019()
positions = [escape_seqs[seq][0]['pos'] for seq in escape_seqs]
min_pos, max_pos = min(positions), max(positions)
analyze_semantics(
args,
model,
vocabulary,
seq_to_mutate,
escape_seqs,
min_pos=min_pos,
max_pos=max_pos,
prob_cutoff=0.,
beta=1.,
plot_acquisition=True,
)
if args.combfit:
from combinatorial_fitness import load_haddox2018
tprint('Haddox et al. 2018...')
wt_seqs, seqs_fitness = load_haddox2018()
strains = sorted(wt_seqs.keys())
for strain in strains:
analyze_comb_fitness(args,
model,
vocabulary,
strain,
wt_seqs[strain],
seqs_fitness,
prob_cutoff=0.,
beta=1.)
def fitness_evcouplings(virus):
if virus == 'h1':
from combinatorial_fitness import load_doud2016
strains, seqs_fitness = load_doud2016()
strain = 'h1'
train_fname = 'target/flu/clusters/all_h1.fasta'
energy_fname = ('target/flu/evcouplings/flu_h1/mutate/'
'flu_h1_single_mutant_matrix.csv')
anchor_id = ('gb:LC333185|ncbiId:BBB04702.1|UniProtKB:-N/A-|'
'Organism:Influenza')
elif virus == 'bf520':
from combinatorial_fitness import load_haddox2018
strains, seqs_fitness = load_haddox2018()
strain = 'BF520'
train_fname = 'target/hiv/clusters/all_BF520.fasta'
energy_fname = ('target/hiv/evcouplings/hiv_bf520/mutate/'
'hiv_bf520_single_mutant_matrix.csv')
anchor_id = 'A1.KE.1994.BF520.W14M.C2.KX168094'
elif virus == 'bg505':
from combinatorial_fitness import load_haddox2018
strains, seqs_fitness = load_haddox2018()
strain = 'BG505'
train_fname = 'target/hiv/clusters/all_BG505.fasta'
energy_fname = ('target/hiv/evcouplings/hiv_env/mutate/'
'hiv_env_single_mutant_matrix.csv')
anchor_id = 'A1.KE.-.BG505_W6M_ENV_C2.DQ208458'
else:
raise ValueError('invalid option {}'.format(virus))
anchor = None
for idx, record in enumerate(SeqIO.parse(train_fname, 'fasta')):
if record.id == anchor_id:
anchor = str(record.seq).replace('-', '')
assert (anchor is not None)
pos_aa_score_epi = {}
pos_aa_score_ind = {}
with open(energy_fname) as f:
f.readline()
for line in f:
fields = line.rstrip().split(',')
pos = int(fields[2]) - 1
orig, mut = fields[3], fields[4]
assert (anchor[pos] == orig)
pos_aa_score_epi[(pos, mut)] = float(fields[7])
pos_aa_score_ind[(pos, mut)] = float(fields[8])
mutations = [
mut_seq for mut_seq, strain_i in seqs_fitness if strain_i == strain
]
mut_scores_epi, mut_scores_ind = [], []
fitnesses = []
for mut_idx, mutation in enumerate(mutations):
mutation = mutation.replace('-', '')
if mutation == anchor:
continue
didx = [c1 != c2 for c1, c2 in zip(anchor, mutation)].index(True)
if (didx, mutation[didx]) in pos_aa_score_epi:
mut_scores_epi.append(pos_aa_score_epi[(didx, mutation[didx])])
mut_scores_ind.append(pos_aa_score_ind[(didx, mutation[didx])])
else:
mut_scores_epi.append(0)
mut_scores_ind.append(0)
fitnesses.append(seqs_fitness[(mutation, strain)][0]['preference'])
print_result(mut_scores_epi, fitnesses, virus + ' (EVcouplings epistatic)')
print_result(mut_scores_ind, fitnesses,
virus + ' (EVcouplings independent)')