def interpolate_null(geneset_size_enrichments, size_skip, go_size):
# find range
size_less = go_size
size_more = go_size
for d in range(size_skip + 1):
if not size_less in geneset_size_enrichments:
size_less -= 1
if not size_more in geneset_size_enrichments:
size_more += 1
# compute interpolation weights
max_dist = 1 + max(go_size - size_less, size_more - go_size)
w_less = float(max_dist - (go_size - size_less))
w_more = float(max_dist - (size_more - go_size))
# compute mean, sd
mean = (w_less * util.mean(geneset_size_enrichments[size_less]) + w_more *
util.mean(geneset_size_enrichments[size_more])) / (w_less + w_more)
sd = (w_less * util.sd(geneset_size_enrichments[size_less]) + w_more *
util.sd(geneset_size_enrichments[size_more])) / (w_less + w_more)
return mean, sd
def read_trip(data):
"""
The main loop is ugly, I know, but I'm doing everything in the same place for performance reasons.
"""
vectors = read_lines(data)
stats = {}
distances = []
accelerations = []
roundto1 = partial(round, ndigits=1)
total_distance = 0.0
resultant_vector = (0, 0)
total_acceleration = 0.0
total_average_speed = 0.0
for i in range(1, len(vectors)):
distance = roundto1(edist(vectors[i], vectors[i - 1]))
distances.append(distance)
total_distance += distance
resultant_vector = (resultant_vector[0] + vectors[i][0],
resultant_vector[1] + vectors[i][1])
average_speed = total_distance / i
total_average_speed += average_speed
acceleration = distance - distances[-1]
accelerations.append(acceleration)
total_acceleration += acceleration
# Drop origin.
#vectors = vectors[1:]
n = len(vectors)
# ?
accelerations = [0] + accelerations
# Calculate the angle in degrees between the final vector and the x-axis. Use atan2
# because it knows about components signal and thus generates correct angles.
# PS. We gotta have a better way to do this.
final_angle = np.digitize(
[degrees(atan2(resultant_vector[1], resultant_vector[0]))],
np.linspace(0, 360, 13))[0]
mean_inst_speed = total_distance / n
sd_inst_speed = sd(distances, mean_inst_speed)
mean_average_speed = total_average_speed / n
sd_average_speed = sd(distances, mean_average_speed)
mean_acceleration = total_acceleration / n
sd_acceleration = sd(accelerations, mean_acceleration)
# Bin and normalize instant speeds frequencies to transform them into feature vector.
distances = np.array(distances)
bins = np.linspace(distances.min(), distances.max(), 80)
digitized = np.digitize(distances, bins)
counts = np.bincount(digitized)
spread = float(counts.max() - counts.min())
counts = (counts - counts.min()) / spread
for n, d in enumerate(counts):
s = 'speed%d' % (n)
stats[s] = d
stats['distance'] = total_distance
stats['points'] = n
stats['mean_inst_speed'] = mean_inst_speed
stats['sd_inst_speed'] = sd_inst_speed
stats['mean_avg_speed'] = mean_average_speed
stats['sd_avg_speed'] = sd_average_speed
stats['mean_acceleration'] = mean_acceleration
stats['sd_acceleration'] = sd_acceleration
stats['final_angle'] = final_angle
# WARNUNG: experimental!
#stats['vars1'] = var_inst_speed + var_average_speed
#stats['vars2'] = var_inst_speed + var_acceleration
#stats['vars3'] = var_average_speed + var_acceleration
#stats['vars4'] = var_inst_speed + var_average_speed + var_acceleration
return stats