def load_lee2019():
fname = 'data/influenza/escape_lee2019/Perth2009_H3_HA.fa'
for record in SeqIO.parse(fname, 'fasta'):
seq = record.seq
break
seqs_escape = {}
fname = 'data/influenza/escape_lee2019/avg_sel_tidy.csv'
with open(fname) as f:
f.readline() # Consume header.
for line in f:
fields = line.rstrip().split(',')
significant = fields[14] == 'True'
pos = int(fields[13])
assert (seq[pos] == fields[5])
escaped = seq[:pos] + fields[6] + seq[pos + 1:]
assert (len(seq) == len(escaped))
if escaped not in seqs_escape:
seqs_escape[escaped] = []
if '-age-' in fields[0]:
species = 'human'
elif 'ferret-' in fields[0]:
species = 'ferret'
else:
species = 'antibody'
seqs_escape[escaped].append({
'abs_diff_selection': float(fields[11]),
'antibody': fields[1],
'species': species,
'significant': significant,
})
return seq, seqs_escape
def load_greaney2020(survival_cutoff=0.3,
binding_cutoff=-2.35, expr_cutoff=-1.5):
seq = SeqIO.read('data/cov/cov2_spike_wt.fasta', 'fasta').seq
sig_sites = set()
with open('data/cov/greaney2020cov2/significant_escape_sites.csv') as f:
f.readline()
for line in f:
fields = line.rstrip().split(',')
sig_sites.add(int(fields[1]) - 1)
binding = {}
with open('data/cov/starr2020cov2/single_mut_effects.csv') as f:
f.readline()
for line in f:
fields = line.rstrip().split(',')
pos = float(fields[1]) - 1
aa_orig = fields[2].strip('"')
aa_mut = fields[3].strip('"')
if aa_mut == '*':
continue
if fields[8] == 'NA':
score = float('-inf')
else:
score = float(fields[8])
if fields[11] == 'NA':
expr = float('-inf')
else:
expr = float(fields[11])
binding[(pos, aa_orig, aa_mut)] = score, expr
seqs_escape = {}
with open('data/cov/greaney2020cov2/escape_fracs.csv') as f:
f.readline() # Consume header.
for line in f:
fields = line.rstrip().split(',')
antibody = fields[2]
escape_frac = float(fields[10])
aa_orig = fields[5]
aa_mut = fields[6]
pos = int(fields[4]) - 1
assert(seq[pos] == aa_orig)
escaped = seq[:pos] + aa_mut + seq[pos + 1:]
assert(len(seq) == len(escaped))
if escaped not in seqs_escape:
seqs_escape[escaped] = []
significant = (
escape_frac >= survival_cutoff and
# Statements below should always be true with defaults.
binding[(pos, aa_orig, aa_mut)][0] >= binding_cutoff and
binding[(pos, aa_orig, aa_mut)][1] >= expr_cutoff
)
seqs_escape[escaped].append({
'pos': pos,
'frac_survived': escape_frac,
'antibody': antibody,
'significant': significant,
})
return seq, seqs_escape
def load_baum2020():
seq = SeqIO.read('data/cov/cov2_spike_wt.fasta', 'fasta').seq
muts = set([
'K417E', 'K444Q', 'V445A', 'N450D', 'Y453F', 'L455F',
'E484K', 'G485D', 'F486V', 'F490L', 'F490S', 'Q493K',
'H655Y', 'R682Q', 'R685S', 'V687G', 'G769E', 'Q779K',
'V1128A',
])
AAs = [
'A', 'R', 'N', 'D', 'C', 'Q', 'E', 'G', 'H',
'I', 'L', 'K', 'M', 'F', 'P', 'S', 'T', 'W',
'Y', 'V',
]
seqs_escape = {}
for idx in range(len(seq)):
for aa in AAs:
if aa == seq[idx]:
continue
mut_seq = seq[:idx] + aa + seq[idx+1:]
mut_str = '{}{}{}'.format(seq[idx], idx + 1, aa)
if mut_seq not in seqs_escape:
seqs_escape[mut_seq] = []
seqs_escape[mut_seq].append({
'mutation': mut_str,
'significant': mut_str in muts,
})
return seq, seqs_escape
def load_haddox2018():
strain_names = ['BF520', 'BG505']
strains = {}
seqs_fitness = {}
for strain in strain_names:
wt_seq = translate(
SeqIO.read(
'data/hiv/fitness_haddox2018/'
'{}_env.fasta'.format(strain), 'fasta').seq).rstrip('*')
strains[strain] = wt_seq
fname = 'data/hiv/fitness_haddox2018/{}_to_HXB2.csv'.format(strain)
pos_map = {}
with open(fname) as f:
f.readline() # Consume header.
for line in f:
fields = line.rstrip().split(',')
pos_map[fields[1]] = (fields[2], int(fields[0]) - 1)
fname = ('data/hiv/fitness_haddox2018/{}_avgprefs.csv'.format(strain))
with open(fname) as f:
mutants = f.readline().rstrip().split(',')[1:]
for line in f:
fields = line.rstrip().split(',')
orig, pos = pos_map[fields[0]]
assert (wt_seq[int(pos)] == orig)
preferences = [float(field) for field in fields[1:]]
assert (len(mutants) == len(preferences))
for mut, pref in zip(mutants, preferences):
mutable = [aa for aa in wt_seq]
mutable[pos] = mut
mut_seq = ''.join(mutable)
if (mut_seq, strain) not in seqs_fitness:
seqs_fitness[(mut_seq, strain)] = [{
'strain':
strain,
'fitnesses': [pref],
'preferences': [pref],
'wildtype':
wt_seq,
'mut_pos': [pos],
}]
else:
seqs_fitness[(mut_seq,
strain)][0]['fitnesses'].append(pref)
seqs_fitness[(mut_seq,
strain)][0]['preferences'].append(pref)
for fit_key in seqs_fitness:
seqs_fitness[fit_key][0]['fitness'] = np.median(
seqs_fitness[fit_key][0]['fitnesses'])
seqs_fitness[fit_key][0]['preference'] = np.median(
seqs_fitness[fit_key][0]['preferences'])
return strains, seqs_fitness
def load_rhee2004(drug_type="PI"):
assert drug_type in ("PI", "NRTI", "NNRTI"), "Invalid drug_type"
# load wt sequence
base_path = "/afs/csail.mit.edu/u/a/andytso/meng/viral-mutation/data/hiv/escape_rhee2004/"
if drug_type == "PI":
wt_fpath = base_path + "hiv1-pr-wt.fasta"
else:
wt_fpath = base_path + "hiv1-rt-wt.fasta"
records = list(SeqIO.parse(wt_fpath, "fasta"))
assert len(records) == 1, "Expecting single wt sequence"
wt_seq = records[0].seq
# load mutations
df = (pd.read_csv(base_path +
"drug_resistance/{}.csv".format(drug_type.lower())).
set_index("Mutation Patterns"))
# filter out multi-mutations and ignore None
df = df[~df.index.str.contains(",")]
df = df[df.index != "None"]
# Stanford mutation indices found at
# https://hivdb.stanford.edu/pages/genotype-phenotype.html
# are 1-based not 0-based
offset = 1
# get the single-residue mutated sequence
def mutate(seq, mutation):
i = int(mutation[:-1])
mutate_to = mutation[-1]
mut_seq = seq.tomutable()
mut_seq[i - offset] = mutate_to
return mut_seq.toseq()
# build seqs_escape where each mutation has a list of entries with fields
# mutation, drug, fold_change, significant
seqs_escape = {}
for mutation, row in df.iterrows():
if mutation == "None":
seq = wt_seq
else:
seq = mutate(wt_seq, mutation)
seqs_escape[seq] = []
for drug in row.index[1:-2]:
seqs_escape[seq].append({
"pos": int(mutation[:-1]) - offset,
"drug": drug,
"resistance fold change": row[drug],
"significant": drug in row["Drugs Escaped"]
})
return wt_seq, seqs_escape
def load_dingens2019(survival_cutoff=0.11):
pos_map = {}
with open('data/hiv/escape_dingens2019/BG505_to_HXB2.csv') as f:
f.readline() # Consume header.
for line in f:
fields = line.rstrip().split(',')
pos_map[fields[1]] = int(fields[0]) - 1
fname = 'data/hiv/escape_dingens2019/Env_protalign_manualeditAD.fasta'
for record in SeqIO.parse(fname, 'fasta'):
if record.description == 'BG505':
seq = record.seq
break
seqs_escape = {}
antibodies = [
'101074',
'10E8',
'3BNC117-101074-pool',
'3BNC117',
'PG9',
'PGT121',
'PGT145',
'PGT151',
'VRC01',
'VRC34',
]
for antibody in antibodies:
fname = ('data/hiv/escape_dingens2019/FileS4/'
'fracsurviveaboveavg/{}.csv'.format(antibody))
with open(fname) as f:
f.readline() # Consume header.
for line in f:
fields = line.rstrip().split(',')
frac_survived = float(fields[3])
pos = pos_map[fields[0]]
assert (seq[pos] == fields[1])
escaped = seq[:pos] + fields[2] + seq[pos + 1:]
assert (len(seq) == len(escaped))
if escaped not in seqs_escape:
seqs_escape[escaped] = []
seqs_escape[escaped].append({
'pos':
pos,
'frac_survived':
frac_survived,
'antibody':
antibody,
'significant':
frac_survived > survival_cutoff,
})
return seq, seqs_escape
def load_doud2018(survival_cutoff=0.05):
pos_map = {}
with open('data/influenza/escape_doud2018/pos_map.csv') as f:
f.readline() # Consume header.
for line in f:
fields = line.rstrip().split(',')
pos_map[fields[1]] = int(fields[0]) - 1
fname = 'data/influenza/escape_doud2018/WSN1933_H1_HA.fa'
seqs = []
for record in SeqIO.parse(fname, 'fasta'):
seq = record.seq
seqs.append(seq)
seqs_escape = {}
antibodies = [
'C179',
'FI6v3',
'H17L10',
'H17L19',
'H17L7',
'S139',
]
for antibody in antibodies:
fname = ('data/influenza/escape_doud2018/' +
'medianfracsurvivefiles/' +
'antibody_{}_median.csv'.format(antibody))
with open(fname) as f:
f.readline() # Consume header.
for line in f:
fields = line.rstrip().split(',')
frac_survived = float(fields[3])
pos = pos_map[fields[0]]
if seq[pos] != fields[1]:
print((seq[pos], fields[1], pos))
assert (seq[pos] == fields[1])
escaped = seq[:pos] + fields[2] + seq[pos + 1:]
assert (len(seq) == len(escaped))
if escaped not in seqs_escape:
seqs_escape[escaped] = []
seqs_escape[escaped].append({
'frac_survived':
frac_survived,
'antibody':
antibody,
'significant':
frac_survived > survival_cutoff,
})
return seq, seqs_escape
def load_doud2016():
strain = 'h1'
fname = 'data/influenza/escape_doud2018/WSN1933_H1_HA.fa'
wt_seq = SeqIO.read(fname, 'fasta').seq
seqs_fitness = {}
fname = ('data/influenza/fitness_doud2016/'
'Supplemental_File_2_HApreferences.txt')
with open(fname) as f:
muts = f.readline().rstrip().split()[4:]
for line in f:
fields = line.rstrip().split()
pos = int(fields[0]) - 1
orig = fields[1]
assert (wt_seq[pos] == orig)
data = [float(field) for field in fields[3:]]
assert (len(muts) == len(data))
for mut, pref in zip(muts, data):
mutable = [aa for aa in wt_seq]
assert (mut.startswith('PI_'))
mutable[pos] = mut[-1]
mut_seq = ''.join(mutable)
assert (len(mut_seq) == len(wt_seq))
if (mut_seq, strain) not in seqs_fitness:
seqs_fitness[(mut_seq, strain)] = [{
'strain': strain,
'fitnesses': [pref],
'preferences': [pref],
'wildtype': wt_seq,
'mut_pos': [pos],
}]
else:
seqs_fitness[(mut_seq,
strain)][0]['fitnesses'].append(pref)
seqs_fitness[(mut_seq,
strain)][0]['preferences'].append(pref)
for fit_key in seqs_fitness:
seqs_fitness[fit_key][0]['fitness'] = np.median(
seqs_fitness[fit_key][0]['fitnesses'])
seqs_fitness[fit_key][0]['preference'] = np.median(
seqs_fitness[fit_key][0]['preferences'])
return {strain: wt_seq}, seqs_fitness
def load_starr2020():
strain = 'sars_cov_2'
wt_seq = SeqIO.read('data/cov/cov2_spike_wt.fasta', 'fasta').seq
seqs_fitness = {}
with open('data/cov/starr2020cov2/binding_Kds.csv') as f:
f.readline()
for line in f:
fields = line.replace('"', '').rstrip().split(',')
if fields[5] == 'NA':
continue
log10Ka = float(fields[5])
mutants = fields[-2].split()
mutable = [aa for aa in wt_seq]
mut_pos = []
for mutant in mutants:
orig, mut = mutant[0], mutant[-1]
pos = int(mutant[1:-1]) - 1 + 330
assert (wt_seq[pos] == orig)
mutable[pos] = mut
mut_pos.append(pos)
mut_seq = ''.join(mutable)
if (mut_seq, strain) not in seqs_fitness:
seqs_fitness[(mut_seq, strain)] = [{
'strain': strain,
'fitnesses': [log10Ka],
'preferences': [log10Ka],
'wildtype': wt_seq,
'mut_pos': mut_pos,
}]
else:
seqs_fitness[(mut_seq, strain)][0]['fitnesses'].append(log10Ka)
seqs_fitness[(mut_seq,
strain)][0]['preferences'].append(log10Ka)
for fit_key in seqs_fitness:
seqs_fitness[fit_key][0]['fitness'] = np.median(
seqs_fitness[fit_key][0]['fitnesses'])
seqs_fitness[fit_key][0]['preference'] = np.median(
seqs_fitness[fit_key][0]['preferences'])
print(len(seqs_fitness))
return {strain: wt_seq}, seqs_fitness
def load_baum2020():
seq = SeqIO.read('data/cov/cov2_spike_wt.fasta', 'fasta').seq
muts = [
'K417E',
'K444Q',
'V445A',
'N450D',
'Y453F',
'L455F',
'E484K',
'G485D',
'F486V',
'F490L',
'F490S',
'Q493K',
'H655Y',
'R682Q',
'R685S',
'V687G',
'G769E',
'Q779K',
'V1128A',
]
seqs_escape = {}
for mut in muts:
aa_orig = mut[0]
aa_mut = mut[-1]
pos = int(mut[1:-1]) - 1
assert (seq[pos] == aa_orig)
escaped = seq[:pos] + aa_mut + seq[pos + 1:]
assert (len(seq) == len(escaped))
if escaped not in seqs_escape:
seqs_escape[escaped] = []
seqs_escape[escaped].append({
'mutation': mut,
'significant': True,
})
return seq, seqs_escape
def load_russ2020(escape_criteria="escape"):
assert escape_criteria in ("escape", "antibiotic-resistance", "combination-resistance"), \
"Invalid escape_criteria"
base_path = "/afs/csail.mit.edu/u/a/andytso/meng/viral-mutation/data/beta_lactamase/escape_russ2020/"
# load wt sequence
wt_fpath = base_path + "ecoli_beta_lactamase_wt.fasta"
records = list(SeqIO.parse(wt_fpath, "fasta"))
assert len(records) == 1, "Expecting single wt sequence"
wt_seq = records[0].seq
# load mutations
df = pd.read_excel(base_path +
"beta_lactamase_inhibitory_concentrations.xlsx",
index_col="Row",
skiprows=[1])
df = df.rename_axis(index="Mutant")
df = df.applymap(lambda x: float(x[1:]) if isinstance(x, str) else x)
# for alignment between russ et al. 2020 indices and that of UniProt
# https://www.nature.com/articles/s41467-020-15666-2
# https://www.uniprot.org/uniprot/P00811#sequences
offset = -15
# get the single-residue mutated sequence
def mutate(seq, mutation, offset=offset):
"""
mutation applied to seq where
mutation is of the form <aa_original><index><aa_mutated>
"""
i = int(mutation[1:-1]) - offset
mutate_from = mutation[0]
mutate_to = mutation[-1]
assert mutate_from == seq[i]
mut_seq = seq.tomutable()
mut_seq[i] = mutate_to
return mut_seq.toseq()
# build seqs_escape where each mutation has a list of entries with fields
# mutation, drug, drug IC50, combination IC50, significant
seqs_escape = {}
for mutation, row in df.iterrows():
if mutation == "WT":
seq = wt_seq
index = None
else:
seq = mutate(wt_seq, mutation)
index = int(mutation[1:-1]) - offset
seqs_escape[seq] = []
for drug in ("PIP", "ATM", "FEP"):
y_label = "{}_AVI".format(drug)
wt_x = df.loc["WT", "{}".format(drug)]
wt_y = df.loc["WT", "{}_AVI".format(drug)]
if escape_criteria == "escape":
is_significant = row[drug] > wt_x and row[y_label] > wt_x
elif escape_criteria == "antibiotic-resistance":
is_significant = row[drug] > wt_x
elif escape_criteria == "combination-resistance":
is_significant = row[y_label] > wt_y
else:
raise ValueError
seqs_escape[seq].append({
"mutation": mutation,
"pos": index,
"drug": drug,
"drug-ic50": row[drug],
"combination-ic50": row[y_label],
"significant": is_significant
})
return wt_seq, seqs_escape
def load_wu2020():
mut_pos = [156, 158, 159, 190, 193, 196]
offset = 16 # Amino acids in prefix.
mut_pos = [pos - 1 + offset for pos in mut_pos]
names = [
'HK68',
'Bk79',
'Bei89',
'Mos99',
'Bris07L194',
'NDako16',
]
wildtypes = [
'KGSESV',
'EESENV',
'EEYENV',
'QKYDST',
'HKFDFA',
'HNSDFA',
]
# Load full wildtype sequences.
wt_seqs = {}
fname = 'data/influenza/fitness_wu2020/wildtypes.fa'
for record in SeqIO.parse(fname, 'fasta'):
strain_idx = names.index(record.description)
wt = wildtypes[strain_idx]
for aa, pos in zip(wt, mut_pos):
assert (record.seq[pos] == aa)
wt_seqs[names[strain_idx]] = record.seq
# Load mutants.
seqs_fitness = {}
fname = 'data/influenza/fitness_wu2020/data_pref.tsv'
with open(fname) as f:
f.readline()
for line in f:
fields = line.rstrip().split('\t')
mut, strain, fitness, preference = fields
if strain == 'Bris07P194':
continue
if strain == 'Bris07':
strain = 'Bris07L194'
fitness = float(preference)
preference = float(preference)
strain_idx = names.index(strain)
wt = wildtypes[strain_idx]
full_seq = wt_seqs[strain]
mutable = [aa for aa in full_seq]
for aa_wt, aa, pos in zip(wt, mut, mut_pos):
assert (mutable[pos] == aa_wt)
mutable[pos] = aa
mut_seq = ''.join(mutable)
if (mut_seq, strain) not in seqs_fitness:
seqs_fitness[(mut_seq, strain)] = []
seqs_fitness[(mut_seq, strain)].append({
'strain': strain,
'fitness': fitness,
'preference': preference,
'wildtype': full_seq,
'mut_pos': mut_pos,
})
return wt_seqs, seqs_fitness