def main():
filename = '../alignments/Q92481'
position = 72
changed_pattern = 'R'
import load_alignments
import uniquify_alignments
import check_mutation_position
import correct_alignment_position
proteins = load_alignments.do(filename)
sorted_list = sorted(proteins, key=itemgetter('match_percentage'))
proteins = sorted_list
proteins.reverse()
proteins = uniquify_alignments.do(proteins)
query, prots = util.fetch_query_protein_in_alignments(proteins)
mod_position = correct_alignment_position.do(query['alignment'], position)
pattern = query['alignment'][mod_position]
patterns = check_mutation_position.do(
[prot['alignment'] for prot in proteins], mod_position)
print(do(patterns, pattern, changed_pattern))
def do(file1, file2, winsize=1):
# fetch all mutations
mutations = []
with open(file1) as fp:
for line in fp:
parts = line.strip().split(',')
mutations.append([parts[0],parts[1],parts[2],parts[3],parts[4]])
# ofp = open(file2,'w')
for mut in mutations:
# fetch reqd info
protein = mut[0]
position = int(mut[1]) - 1
orig_aa = mut[2]
mut_aa = mut[3]
# if protein not aligned - pass this
alignment_file = '../alignments/%s'%protein
if not os.path.isfile(alignment_file):
continue
print(mut)
# load the aligned sequences
alignments = load_alignments.do(alignment_file)
# remove duplicate sequences in aligned sequences
alignments = uniquify_alignments.do(alignments)
# fetch query sequence
query, alignments = util.fetch_query_protein_in_alignments(alignments)
if len(alignments) == 0:
continue
# fetch actual position of mutation in aligned query sequence
try:
actual_pos = correct_alignment_position.do(query['alignment'], position)
except Exception as e:
print(query['alignment'])
continue
o_win_score = 0
for w in range(winsize+1):
scores = [o_score, m_score, o_ps_score, m_ps_score, o_sw_score, m_sw_score, o_gf_score, m_gf_score]
print(scores)
#ofp.write(','.join([ str(item) for item in list(chain.from_iterable([mut, scores])) ])+'\n')
break
#ofp.close()
def main():
args = sys.argv
if len(args) > 1:
f1 = args[1] # where mutation info is present
f2 = args[2] # where we store the scores
winsize = args[3]
else:
f1 = 'datasets/test/unseen_proteins.csv'
f2 = 'datasets/test/unseen_proteins_freq_scores.csv'
winsize = 3
do(f1, f2, winsize)
if __name__ == '__main__':
main()
def do(file1, file2, nbsize):
mutations = []
with open(file1) as fp:
for line in fp:
mutations.append(line.strip().split(','))
mutations.reverse()
with open(file2,'w') as ofp:
# ---------------------------------------------------------
# write the heading to the output file
heading = ['Protein','Position','Orig_aa','Mut_aa','Label']
# neighbour part
for n in range(1, nbsize+1):
for i in range(n+1):
if n > 3:
maxhd = 2
elif n > 2:
maxhd = 1
else:
maxhd = 0
for k in range(maxhd+1):
# hamming distance k
heading.append('nbwidth:%s srtpos:%s hd:%s orig' %(str(n), str(-i), str(k)))
heading.append('nbwidth:%s srtpos:%s hd:%s mut' %(str(n), str(-i), str(k)))
#print heading
ofp.write(','.join(heading)+'\n')
# ---------------------------------------------------------
# compute scores for mutations
for mut in mutations:
protein = mut[0]
position = int(mut[1]) - 1
orig_aa = mut[2]
mut_aa = mut[3]
alignment_path = '%s/%s' %(aligndir, protein)
if not os.path.exists(alignment_path):
continue
#print(mut)
query_flag = 0 # to capture IndexError in query protein sequence
# fetch alignments, and query protein
alignments = load_alignments.do(alignment_path)
query, alignments = util.fetch_query_protein_in_alignments(alignments)
alignments = util.prune_proteins_list(alignments)
# fetch all sequences
sequences = [a['alignment'] for a in alignments]
query_seq = query['alignment']
# all neighbour scores
scores = []
# go into neighbour mode now
for n in range(1, nbsize+1):
if n > 3:
maxhd = 2
elif n > 2:
maxhd = 1
else:
maxhd = 0
for i in range(n+1):
# i is the relative position of mutation position
# in nb_pos list
start_pos = position - i
# fetch the indices of neighbours for this round
nb_pos = []
try:
for j in range(n+1):
# correct the position in the alignment
cpos = correct_alignment_position.do(query_seq, start_pos+j)
nb_pos.append(cpos)
except Exception:
query_flag = 1
break
# fetch corresponding positions of aligned sequences
seq_parts = []
for seq in sequences:
s = list(seq)
row = []
flag = 0
for pos in nb_pos:
try:
# do correction for 'B' and 'Z' amino acids
if s[pos] == 'B':
row.append('N')
elif s[pos] == 'Z':
row.append('Q')
else:
row.append(s[pos])
except IndexError:
flag = 1
break
if flag:
continue
seq_parts.append(row)
if len(seq_parts) == 0:
query_flag = 1
break
if query_flag:
break
# fetch corresponding positions of query sequence
query_seq_part = []
s = list(query_seq)
for pos in nb_pos:
try:
# do correction for 'B' and 'Z' amino acids
if s[pos] == 'B':
query_seq_part.append('N')
elif s[pos] == 'Z':
query_seq_part.append('Q')
else:
query_seq_part.append(s[pos])
except IndexError:
query_flag = 1
break
if query_flag:
break
# construct the mutated query sequence's parts
mut_query_seq_part = list(query_seq_part)
mut_query_seq_part[i] = mut_aa
for k in range(maxhd+1):
#print('%s %s %s - original' %(str(n),str(i),str(k)))
# compute hd-k orig score
scores.append(calc_hd_score(seq_parts, query_seq_part, i, k))
#print('%s %s %s - mutant' %(str(n),str(i),str(k)))
# compute hd-k mut score
scores.append(calc_hd_score(seq_parts, mut_query_seq_part, i, k))
if query_flag:
break
if query_flag:
continue
#print scores
ofp.write(','.join([ str(item) for item in list(chain.from_iterable([mut, scores])) ])+'\n')
orig_aa = mut[2]
mut_aa = mut[3]
# if protein not aligned - pass this
alignment_file = '../alignments/%s' % protein
if not os.path.isfile(alignment_file):
continue
# load the aligned sequences
alignments = load_alignments.do(alignment_file)
# prune the proteins
alignments = util.prune_proteins_list(alignments)
# fetch query sequence
query, alignments = util.fetch_query_protein_in_alignments(alignments)
if len(alignments) == 0:
continue
# fetch actual position of mutation in aligned query sequence
try:
actual_pos = correct_alignment_position.do(query['alignment'],
position)
except Exception as e:
print(query['alignment'])
continue
# fetch the corresponding sequences
sequences = [a['alignment'] for a in alignments]
# compute frequency of amino acid at desired position
def work(line):
'''This method computes all kinds of scores,
for a given mutation, and returns the computed scores.
Args:
line: which contains mutation info.
<PROTEIN_ID, POSITION, ORIGINAL_AMINO_ACID, MUTANT_AMINO_ACID>
Return Values:
[parts, scores] flattend list
'''
line = line.strip()
parts = line.split(',')
protein = parts[0]
position = int(parts[1]) - 1 # correction for indexing python lists
mut_aa = parts[3]
result = []
print(protein)
for alignment_dir in align_dirs:
alignment_file = os.path.join(alignment_dir, protein)
if not os.path.isfile(alignment_file):
return []
fasta = util.read_sequence('../fasta/%s.fasta' % protein)
proteins = load_alignments.do(
alignment_file) # types of proteins required
# sort & prune the list of proteins
proteins = prune_proteins_list(proteins)
# fetch the record pertaining to the query protein
query, p = util.fetch_query_protein_in_alignments(
proteins) # p does not contain query sequence
prots = [prot["id"] for prot in proteins]
types = [prot["type"] for prot in proteins]
match_percents = [prot["match_percentage"] for prot in proteins]
alignments = [prot["alignment"] for prot in proteins]
proteins = prots #replacing proteins dicitonary with only ids
result.append(len(prots) + 1)
orig_aa = fasta[position]
# correct the position of query protein wrt to alignment
mod_pos = correct_alignment_position.do(query['alignment'], position)
aa = check_mutation_position.do(alignments, mod_pos)
# calculate shannon entropy score w/o sequence weights
result.append(
shannon_entropy_score.do(list(alignments), query['alignment'],
mod_pos, orig_aa, mut_aa))
# shannon entropy score with sequence weights
result.append(
shannon_entropy_score.do(list(alignments), query['alignment'],
mod_pos, orig_aa, mut_aa, 1))
# calculate von-neumann entropy score
result.append(von_neumann_entropy_score.do(list(aa), orig_aa, mut_aa))
# calculate relative entropy score
#result.append(relative_entropy_score.do(list(aa), orig_aa, mut_aa))
# calculate jensen-shannon divergence score
#result.append(jensen_shannon_divergence_score(list(aa), orig_aa, mut_aa))
# calculate sum-of-pairs scores
result.append(
sum_of_pairs_score.do(list(alignments), query['alignment'],
mod_pos, orig_aa, mut_aa, 0)) # wo seq wg
result.append(
sum_of_pairs_score.do(list(alignments), query['alignment'],
mod_pos, orig_aa, mut_aa, 1)) # w seq wg
# return mutation information and scores for recording in a file
return [str(item) for item in list(chain.from_iterable([parts, result]))]
def do(file1, file2):
# fetch all mutations
mutations = []
with open(file1) as fp:
for line in fp:
parts = line.strip().split(',')
mutations.append(
[parts[0], parts[1], parts[2], parts[3], parts[4]])
ofp = open(file2, 'w')
for mut in mutations:
# fetch reqd info
protein = mut[0]
position = int(mut[1]) - 1
orig_aa = mut[2]
mut_aa = mut[3]
# if protein not aligned - pass this
alignment_file = '../alignments/%s' % protein
if not os.path.isfile(alignment_file):
continue
print(mut)
# load the aligned sequences
alignments = load_alignments.do(alignment_file)
# prune the proteins
alignments = util.prune_proteins_list(alignments)
# fetch query sequence
query, alignments = util.fetch_query_protein_in_alignments(alignments)
if len(alignments) == 0:
continue
# fetch actual position of mutation in aligned query sequence
try:
actual_pos = correct_alignment_position.do(query['alignment'],
position)
except Exception as e:
print(query['alignment'])
continue
# fetch the corresponding sequences
sequences = [a['alignment'] for a in alignments]
# compute frequency of amino acid at desired position
aa = check_mutation_position.do([k['alignment'] for k in alignments],
actual_pos)
# compute simple frequency of original & mutant amino acid
o_score, m_score = simple_frequency_score.do(list(aa), orig_aa, mut_aa)
# compute score using pseudo-counts in order to account for missing aa
o_ps_score, m_ps_score = pseudo_count_score.do(list(aa), orig_aa,
mut_aa)
# compute simple sequence-weighted frequency score
sequence_weights = landgraf_sequence_weights.do(
[a['alignment'] for a in alignments])
o_sw_score, m_sw_score = simple_frequency_score.do(
list(aa), orig_aa, mut_aa, sequence_weights)
# using gap frequencies
o_gf_score, m_gf_score = gapped_frequency_score.do(
list(aa), orig_aa, mut_aa, sequence_weights)
# calculate shannon entropy score w/o sequence weights
shannon = shannon_entropy_score.do(list(sequences), query['alignment'],
actual_pos, orig_aa, mut_aa)
# shannon entropy score with sequence weights
shannon_weighted = shannon_entropy_score.do(list(sequences),
query['alignment'],
actual_pos, orig_aa,
mut_aa, 1)
# calculate von-neumann entropy score
von_neumann_score = von_neumann_entropy_score.do(
list(aa), orig_aa, mut_aa)
# calculate sum-of-pairs scores
sop = sum_of_pairs_score.do(list(sequences), query['alignment'],
actual_pos, orig_aa, mut_aa,
0) # wo seq wg
sop_wg = sum_of_pairs_score.do(list(sequences), query['alignment'],
actual_pos, orig_aa, mut_aa,
1) # w seq wg
# append all scores together and write to file
scores = [
o_score, m_score, o_ps_score, m_ps_score, o_sw_score, m_sw_score,
o_gf_score, m_gf_score, shannon, shannon_weighted,
von_neumann_score, sop, sop_wg
]
#print(scores)
ofp.write(','.join(
[str(item)
for item in list(chain.from_iterable([mut, scores]))]) + '\n')
#break
ofp.close()
def do(file1, file2, winsize):
mutations = []
with open(file1) as fp:
for line in fp:
mutations.append(line.strip().split(','))
with open(file2, 'w') as ofp:
for mut in mutations:
print mut
flag = 0
protein = mut[0]
position = int(mut[1]) - 1
orig_aa = mut[2]
mut_aa = mut[3]
fasta_seq = util.read_sequence('../fasta/%s.fasta' % protein)
alfile = '../alignments/%s' % protein
if not os.path.isfile(alfile):
continue
proteins = load_alignments.do(alfile)
proteins = util.prune_proteins_list(proteins)
query, proteins = util.fetch_query_protein_in_alignments(proteins)
query_seq = query['alignment']
alignments = [a['alignment'] for a in proteins]
try:
sequence_weights = landgraf_sequence_weights.do(alignments)
except Exception as e:
print str(e)
flag = 1
if flag:
continue
scores = []
for w in range(winsize + 1): # 0,1,2,3 for winsize=3, hence the +1
try:
# what score to use?
if w == 0:
mod_pos = correct_alignment_position.do(
query_seq, position)
aa = check_mutation_position.do(alignments, mod_pos)
o, m = gapped_frequency_score.do(
list(aa), orig_aa, mut_aa,
sequence_weights) # mod_pos
scores.append(m)
else:
mod_pos = correct_alignment_position.do(
query_seq, position - w)
aa = check_mutation_position.do(alignments, mod_pos)
aa_in_fasta = fasta_seq[position - w]
o, m = gapped_frequency_score.do(
list(aa), aa_in_fasta, mut_aa,
sequence_weights) # mod_pos
scores.append(
o
) # left neighbour at position w from mutation position
mod_pos = correct_alignment_position.do(
query_seq, position + w)
aa = check_mutation_position.do(alignments, mod_pos)
aa_in_fasta = fasta_seq[position + w]
o, m = gapped_frequency_score.do(
list(aa), aa_in_fasta, mut_aa,
sequence_weights) # mod_pos
scores.append(
o
) # right neighbour at position w from mutation position
except Exception:
flag = 1
break
if flag:
continue
#print(scores)
ofp.write(','.join([
str(item) for item in list(chain.from_iterable([mut, scores]))
]) + '\n')
def do(file1, file2):
# fetch all mutations
mutations = []
with open(file1) as fp:
for line in fp:
parts = line.strip().split(',')
mutations.append(
[parts[0], parts[1], parts[2], parts[3], parts[4]])
# ofp = open(file2,'w')
for mut in mutations:
# fetch reqd info
protein = mut[0]
position = int(mut[1]) - 1
orig_aa = mut[2]
mut_aa = mut[3]
# if protein not aligned - pass this
alignment_file = '../alignments/%s' % protein
if not os.path.isfile(alignment_file):
continue
print(mut)
# load the aligned sequences
alignments = load_alignments.do(alignment_file)
# remove duplicate sequences in aligned sequences
alignments = uniquify_alignments.do(alignments)
# fetch query sequence
query, alignments = util.fetch_query_protein_in_alignments(alignments)
if len(alignments) == 0:
continue
# fetch actual position of mutation in aligned query sequence
try:
actual_pos = correct_alignment_position.do(query['alignment'],
position)
except Exception as e:
print(query['alignment'])
continue
# compute frequency of amino acid at desired position
aa = check_mutation_position.do([k['alignment'] for k in alignments],
actual_pos)
# compute simple frequency of original & mutant amino acid
o_score, m_score = simple_frequency_score.do(aa, orig_aa, mut_aa)
# compute score using pseudo-counts in order to account for missing aa
o_ps_score, m_ps_score = pseudo_count_score.do(aa, orig_aa, mut_aa)
# compute simple sequence-weighted frequency score
sequence_weights = landgraf_sequence_weights.do(
[a['alignment'] for a in alignments])
o_sw_score, m_sw_score = simple_frequency_score.do(
aa, orig_aa, mut_aa, sequence_weights)
# using gap frequencies
o_gf_score, m_gf_score = gapped_frequency_score.do(
aa, orig_aa, mut_aa, sequence_weights)
# append all scores together and write to file
scores = [
o_score, m_score, o_ps_score, m_ps_score, o_sw_score, m_sw_score,
o_gf_score, m_gf_score
]
print(scores)
# ofp.write(','.join([ str(item) for item in list(chain.from_iterable([mut, scores])) ])+'\n')
break