def get_best_hamming_enrichment_from_counts(w, P, N, pos_counts, neg_counts, print_dists=False):
"""
Find the most enriched Hamming distance for given counts.
Returns (best_dist, best_log_pvalue, best_p, best_n)
"""
# get cumulative counts
cum_pos_counts = copy.copy(pos_counts)
cum_neg_counts = copy.copy(neg_counts)
for i in range(1, len(pos_counts)):
cum_pos_counts[i] += cum_pos_counts[i-1]
cum_neg_counts[i] += cum_neg_counts[i-1]
# compute hypergeometric enrichment at each distance and save best
best_dist = w
best_log_pvalue = 1 # infinity
best_p = 0
best_n = 0
for i in range(len(pos_counts)):
p = cum_pos_counts[i]
n = cum_neg_counts[i]
log_pvalue = getLogFETPvalue(p, P, n, N, best_log_pvalue)
if log_pvalue < best_log_pvalue:
best_log_pvalue = log_pvalue
best_dist = i
best_p = p
best_n = n
if print_dists:
pv_string = sprint_logx(log_pvalue, 1, _pv_format)
print "d %d : %d %d %d %d %s" % (i, p, P, n, N, pv_string)
return best_dist, best_log_pvalue, best_p, best_n
def get_best_hamming_enrichment_from_counts(w, P, N, pos_counts, neg_counts, print_dists=False):
"""
Find the most enriched Hamming distance for given counts.
Returns (best_dist, best_log_pvalue, best_p, best_n)
"""
# get cumulative counts
cum_pos_counts = copy.copy(pos_counts)
cum_neg_counts = copy.copy(neg_counts)
for i in range(1, len(pos_counts)):
cum_pos_counts[i] += cum_pos_counts[i-1]
cum_neg_counts[i] += cum_neg_counts[i-1]
# compute hypergeometric enrichment at each distance and save best
best_dist = w
best_log_pvalue = 1 # infinity
best_p = 0
best_n = 0
for i in range(len(pos_counts)):
p = cum_pos_counts[i]
n = cum_neg_counts[i]
log_pvalue = getLogFETPvalue(p, P, n, N, best_log_pvalue)
if log_pvalue < best_log_pvalue:
best_log_pvalue = log_pvalue
best_dist = i
best_p = p
best_n = n
if print_dists:
pv_string = sprint_logx(log_pvalue, 1, _pv_format)
print "d %d : %d %d %d %d %s" % (i, p, P, n, N, pv_string)
return best_dist, best_log_pvalue, best_p, best_n
def fit_fisher(p, P, n, N):
#pvalue_threshold = 0.5
pvalue_threshold = False # Modification to allow p-value above 0.5
return getLogFETPvalue(p, P, n, N, pvalue_threshold)