def calc_markov_length_end(raw_data, P, Q, actual):
pos_data = gene_predictor.predict_by_threshold(raw_data)
pos_data = pos_data[0] + pos_data[1]
v = []
for orf in pos_data:
start = orf[0]
end = orf[1]
seq = raw_data[start-1:end]
length = len(seq)
markov = gene_predictor.markov_score(P, Q, seq)
is_valid = end in actual
v.append((length, markov, is_valid))
return v
def generate_roc_length(actual, raw_data):
fp = []
tp = []
end_pos = []
for i in range(50, 1410, 10):
preds = gene_predictor.predict_by_threshold(raw_data, i)
true_positives, false_positives, false_negatives = gene_predictor.check_predictions(preds[1], actual)
tru_pos_rate = true_positives / len(actual)
false_pos_rate = false_positives / (len(preds[0]) + len(preds[1]) - len(actual))
tp.append(tru_pos_rate)
fp.append(false_pos_rate)
return tp, fp
def generate_roc_markov(actual, raw_data, P, Q):
fp = []
tp = []
all_orfs = gene_predictor.predict_by_threshold(raw_data)
all_orfs = all_orfs[0] + all_orfs[1]
for markov_score in range(100):
preds = []
total = 0
for r in all_orfs:
start = r[0] - 1
end = r[1]
seq = raw_data[start:end]
mm = gene_predictor.markov_score(P, Q, seq)
if mm > (float(markov_score)/5):
preds.append(r)
total += 1
true_positives, false_positives, false_negatives = gene_predictor.check_predictions(preds, actual)
tru_pos_rate = true_positives / len(actual)
false_pos_rate = false_positives / (total - len(actual))
tp.append(tru_pos_rate)
fp.append(false_pos_rate)
return tp, fp
f.write(str(tp_m[i]) + "," + str(fp_m[i]) + '\n')
with open("length_roc_curve.csv", 'w') as f:
for i in range(len(tp_l)):
f.write(str(tp_l[i]) + "," + str(fp_l[i]) + '\n')
if write_markov_roc_csv:
graphVals(valid_orfs, invalid_orfs, unk_orfs, actual, raw_data, P, Q)
if uh:
v = calc_markov_length_end(raw_data, P, Q, actual)
calc_line(v)
if write_length_csv:
res = []
for i in range(50, 1500, 10):
pred = gene_predictor.predict_by_threshold(raw_data, i)
tru_pos, false_pos, false_neg = gene_predictor.check_predictions(pred[1], actual)
res.append((str(i), str(float(tru_pos)/(tru_pos+false_neg)), str(tru_pos), str(false_pos)))
#print(str(tru_pos) + " " + str(false_pos) + " " + str(false_neg))
f = open('length.csv', 'w')
for point in res:
f.write(point[0] + ', ' + point[1] + ', ' + point[2] + ', ' + point[3] + '\n')
f.close()
if write_markov_csv:
res = []
for markov_score in range(100):
all_orfs = gene_predictor.predict_by_threshold(raw_data)
all_orfs = all_orfs[0] + all_orfs[1]
pred = []