def seq_pairs(S, enzymes, mat_adj=None, conv_dict=None, return_groups=False):
"""
- S: stoichiometric matrix of reactants (-) and products (+)
- enzymes: list of enzymes corresponding to columns in S
"""
mat_adj = mat_adj if mat_adj is not None else adj_matrix(S)
rows, cols = np.where(mat_adj > 0)
#wrestling with where output
rows, cols = [np.array(x)[0] for x in rows, cols]
rowcols = ut.zip_exact(rows, cols)
# dedup, keeping only upper off-diagonal
rowcols = [(row,col) for row,col in rowcols if row < col]
pairs = [(enzymes[row],enzymes[col]) for row,col in rowcols]
# filter out blanks
labeled_pairs = [(x,y) for x,y in pairs if x and y]
if return_groups:
if conv_dict:
return convert_groups_singles(labeled_pairs, conv_dict)
else:
return labeled_pairs
single_pairs = [(xi,yi) for x,y in labeled_pairs
for xi in x.split() for yi in y.split()]
unique_pairs = pu.dedupe(single_pairs)
print "%s total, %s labeled, %s single, %s unique pairs returned" % (len(pairs),
len(labeled_pairs), len(single_pairs), len(unique_pairs))
if conv_dict:
conv_pairs = convert_pairs_singles(unique_pairs, conv_dict)
print "%s converted pairs with 1-1 matches" % len(conv_pairs)
return conv_pairs
else:
return unique_pairs
def pairs_exceeding(elut, skey, thresh):
"""
Doesn't return self-self interactions.
"""
arr_prots = np.array(elut.prots)
if skey == 'apex':
apexes = ApexScores(elut).apex_array
pairs = matching_pairs(apexes, arr_prots)
else: # loading precomputed indices is so far massively slower than this
score_mat, _, new_prots = scorekey_elution(skey, elut, None)
if new_prots is not None:
arr_prots = np.array(new_prots)
rows, cols = np.where(score_mat > thresh)
p1s, p2s = [arr_prots[ids] for ids in rows, cols]
pairs = ut.zip_exact(p1s, p2s)
return pairs
def pairs_exceeding(elut, skey, thresh):
"""
Doesn't return self-self interactions.
"""
arr_prots = np.array(elut.prots)
if skey == 'apex':
apexes = ApexScores(elut).apex_array
pairs = matching_pairs(apexes, arr_prots)
else: # loading precomputed indices is so far massively slower than this
score_mat, _, new_prots = scorekey_elution(skey, elut, None)
if new_prots is not None:
arr_prots = np.array(new_prots)
rows, cols = np.where(score_mat > thresh)
p1s, p2s = [arr_prots[ids] for ids in rows, cols]
pairs = ut.zip_exact(p1s, p2s)
return pairs
def transpose(d, fin, fout):
sys.path.append(d+'/..')
import utils as ut
lines = [l for l in ut.load_tab_file(fin)]
if lines[-1][0].startswith('#'):
#ignore comments, such as last line in spcount output
lines = lines[:-1]
print "skipping last line"
cols = ut.zip_exact(*lines) #zip messes up if these files aren't neat
# _After_ zipping, get rid of the column 1 header--R doesn't like it.
col0list = list(cols[0])
print col0list[0][0]
assert (col0list[0][0] == '#' or col0list[0] == 'Locus') # make sure we're removing what we should be
col0list.remove(col0list[0])
cols[0] = tuple(col0list)
col2title = cols[1][0].lower()
# get rid of the total/descr column
if col2title.find('total') > -1 or col2title.find('descr') > -1:
cols.remove(cols[1])
print "removing second column--extraneous"
ut.write_tab_file(cols, fout)
def combine_ppis_matched(ppisa, ppisb):
return [(pa[0],pa[1],combine_or(pa[2],pb[2]),pa[3])
for pa,pb in ut.zip_exact(ppisa, ppisb)]