def score_sequences(_pair, args):
# Calculate the best possible scores, and divide by the observed scores
id1, id2 = _pair
alb_obj, psi_pred_files, outfile = args
id_regex = "^%s$|^%s$" % (id1, id2)
alb_copy = Alb.make_copy(alb_obj)
Alb.pull_records(alb_copy, id_regex)
observed_score = 0
seq1_best = 0
seq2_best = 0
seq1, seq2 = alb_copy.records()
prev_aa1 = "-"
prev_aa2 = "-"
for aa_pos in range(alb_copy.lengths()[0]):
aa1 = seq1.seq[aa_pos]
aa2 = seq2.seq[aa_pos]
if aa1 != "-":
seq1_best += BLOSUM62[aa1, aa1]
if aa2 != "-":
seq2_best += BLOSUM62[aa2, aa2]
if aa1 == "-" or aa2 == "-":
if prev_aa1 == "-" or prev_aa2 == "-":
observed_score += gap_extend
else:
observed_score += gap_open
else:
observed_score += BLOSUM62[aa1, aa2]
prev_aa1 = str(aa1)
prev_aa2 = str(aa2)
subs_mat_score = ((observed_score / seq1_best) + (observed_score / seq1_best)) / 2
# PSI PRED comparison
num_gaps = 0
ss_score = 0
for row1 in psi_pred_files[id1].itertuples():
if (psi_pred_files[id2]["indx"] == row1.indx).any():
row2 = psi_pred_files[id2].loc[psi_pred_files[id2]["indx"] == row1.indx]
row_score = 0
row_score += 1 - abs(float(row1.coil_prob) - float(row2.coil_prob))
row_score += 1 - abs(float(row1.helix_prob) - float(row2.helix_prob))
row_score += 1 - abs(float(row1.sheet_prob) - float(row2.sheet_prob))
ss_score += row_score / 3
else:
num_gaps += 1
align_len = len(psi_pred_files[id2]) + num_gaps
ss_score /= align_len
final_score = (ss_score * 0.3) + (subs_mat_score * 0.7)
with lock:
with open(outfile, "a") as _ofile:
_ofile.write("\n%s,%s,%s" % (id1, id2, final_score))
return
def score_sequences(seq_pair):
seq1, seq2 = seq_pair.records
id_regex = "^%s$|^%s$" % (seq1.id, seq2.id)
sb_copy = Sb.make_copy(seqbuddy)
Sb.delete_records(sb_copy, id_regex)
sb_copy = Sb.SeqBuddy(sb_copy.records + [seq1, seq2], out_format="gb", alpha=sb_copy.alpha)
alignbuddy = Alb.generate_msa(sb_copy, tool="mafft", params=" --globalpair", quiet=True)
if not in_args.no_msa_trim:
alignbuddy = Alb.trimal(alignbuddy, threshold="gappyout")
alignbuddy = Alb.pull_records(alignbuddy, id_regex)
_score = 0
seq1, seq2 = alignbuddy.records()
prev_aa1 = "-"
prev_aa2 = "-"
for aa_pos in range(alignbuddy.lengths()[0]):
aa1 = seq1.seq[aa_pos]
aa2 = seq2.seq[aa_pos]
if aa1 == "-" or aa2 == "-":
if prev_aa1 == "-" or prev_aa2 == "-":
_score += gap_extend
else:
_score += gap_open
else:
_score += BLOSUM45[aa1, aa2]
prev_aa1 = str(aa1)
prev_aa2 = str(aa2)
return _score
def test_pull_records(key, next_hash, alb_resources, hf):
alignbuddy = alb_resources.get_one(key)
Alb.pull_records(alignbuddy, "α[1-5]$|β[A-M]")
assert hf.buddy2hash(alignbuddy) == next_hash, alignbuddy.write("error_files%s%s" % (next_hash, os.path.sep))
