Hoaxx / app.py
Dmtlant's picture
Update app.py
70a14b2 verified
import streamlit as st
import numpy as np
import matplotlib.pyplot as plt
import random
from scipy.stats import entropy as scipy_entropy
import time
import seaborn as sns
# --- НАСТРОЙКИ ---
seqlen = 60
steps = 120
min_run, max_run = 1, 2
ANGLE_MAP = {'A': 60.0, 'C': 180.0, 'G': -60.0, 'T': -180.0, 'N': 0.0}
bases = ['A', 'C', 'G', 'T']
def find_local_min_runs(profile, min_run=1, max_run=2):
result = []
N = len(profile)
i = 0
while i < N:
run_val = profile[i]
run_length = 1
while i + run_length < N and profile[i + run_length] == run_val:
run_length += 1
if min_run <= run_length <= max_run:
result.append((i, i + run_length - 1, run_val))
i += run_length
return result
def bio_mutate(seq):
r = random.random()
if r < 0.70: # Точечная мутация
idx = random.randint(0, len(seq)-1)
orig = seq[idx]
prob = random.random()
if orig in 'AG':
newbase = 'C' if prob < 0.65 else random.choice(['T', 'C'])
elif orig in 'CT':
newbase = 'G' if prob < 0.65 else random.choice(['A', 'G'])
else:
newbase = random.choice([b for b in bases if b != orig])
seq = seq[:idx] + newbase + seq[idx+1:]
elif r < 0.80: # Инсерция короткого блока
idx = random.randint(0, len(seq)-1)
ins = ''.join(random.choices(bases, k=random.randint(1, 3)))
seq = seq[:idx] + ins + seq[idx:]
if len(seq) > seqlen:
seq = seq[:seqlen]
elif r < 0.90: # Делеция
if len(seq) > 4:
idx = random.randint(0, len(seq)-2)
dell = random.randint(1, min(3, len(seq)-idx))
seq = seq[:idx] + seq[idx+dell:]
else: # Блочная перестановка (инверсия)
if len(seq) > 10:
start = random.randint(0, len(seq)-6)
end = start + random.randint(3,6)
subseq = seq[start:end]
subseq = subseq[::-1]
seq = seq[:start] + subseq + seq[end:]
while len(seq) < seqlen:
seq += random.choice(bases)
if len(seq) > seqlen:
seq = seq[:seqlen]
return seq
def compute_autocorr(profile):
profile = profile - np.mean(profile)
result = np.correlate(profile, profile, mode='full')
result = result[result.size // 2:]
norm = np.max(result) if np.max(result)!=0 else 1
return result[:10]/norm
def compute_entropy(profile):
vals, counts = np.unique(profile, return_counts=True)
p = counts / counts.sum()
return scipy_entropy(p, base=2)
# --- Streamlit интерфейс ---
st.title("🧬 Эволюция ДНК-подобной последовательности")
st.markdown("Модель визуализирует мутации и анализирует структуру последовательности во времени.")
# Функции для улучшения визуализации
def plot_density(runs, ax, steps):
machine_starts = [start for start, _, _ in runs]
machine_ends = [end for _, end, _ in runs]
timeline = np.zeros(steps)
for start, end, _ in runs:
timeline[start:end+1] = 1
sns.lineplot(x=np.arange(steps), y=timeline, ax=ax, color='darkgreen')
ax.set_title("Плотность биомашин по шагам")
ax.set_xlabel("Шаг")
ax.set_ylabel("Плотность биомашин")
def plot_heatmap(runs, ax, steps):
heatmap_matrix = np.zeros((steps, len(runs)))
for idx, (start, end, _) in enumerate(runs):
heatmap_matrix[start:end+1, idx] = 1
sns.heatmap(heatmap_matrix, ax=ax, cmap="YlGnBu", cbar=True)
ax.set_title("Тепловая карта: Распределение биомашин по времени")
ax.set_xlabel("Номер биомашины")
ax.set_ylabel("Шаг")
def update_visualization(step, seq, runs, torsion_profile, stat_bist_counts, acorr, steps, lags_shown, ent):
fig, axs = plt.subplots(5, 1, figsize=(10, 12))
plt.subplots_adjust(hspace=0.45)
axs[0].cla()
axs[1].cla()
axs[2].cla()
axs[3].cla()
axs[4].cla()
axs[0].plot(torsion_profile, color='royalblue', label="Торсионный угол")
for start, end, val in runs:
axs[0].axvspan(start, end, color="red", alpha=0.3)
axs[0].plot(range(start, end+1), torsion_profile[start:end+1], 'ro', markersize=5)
axs[0].set_ylim(-200, 200)
axs[0].set_xlabel("Позиция")
axs[0].set_ylabel("Торсионный угол (град.)")
axs[0].set_title(f"Шаг {step}: {seq}\nЧисло машин: {len(runs)}, энтропия: {ent:.2f}")
axs[0].legend()
plot_density(runs, axs[3], steps)
plot_heatmap(runs, axs[4], steps)
axs[1].plot(stat_bist_counts, '-o', color='crimson', markersize=4)
axs[1].set_xlabel("Шаг")
axs[1].set_ylabel("Число машин")
axs[1].set_ylim(0, max(10, max(stat_bist_counts)+1))
axs[1].set_title("Динамика: число 'биомашин'")
axs[2].bar(np.arange(lags_shown), acorr[:lags_shown], color='teal', alpha=0.7)
axs[2].set_xlabel("Лаг")
axs[2].set_ylabel("Автокорреляция")
axs[2].set_title("Автокорреляция углового профиля (структурность) и энтропия")
axs[2].text(0.70, 0.70, f"Энтропия: {ent:.2f}", transform=axs[2].transAxes)
plot_placeholder.pyplot(fig)
# Кнопка запуска симуляции
if st.button("▶️ Запустить симуляцию"):
seq = ''.join(random.choices(bases, k=seqlen))
stat_bist_counts = []
stat_entropy = []
stat_autocorr = []
plot_placeholder = st.empty()
for step in range(steps):
if step != 0:
seq = bio_mutate(seq)
torsion_profile = np.array([ANGLE_MAP.get(nt, 0.0) for nt in seq])
runs = find_local_min_runs(torsion_profile, min_run, max_run)
stat_bist_counts.append(len(runs))
ent = compute_entropy(torsion_profile)
stat_entropy.append(ent)
acorr = compute_autocorr(torsion_profile)
# Обновление визуализаций
update_visualization(step, seq, runs, torsion_profile, stat_bist_counts, acorr, steps, 6, ent)
time.sleep(0.5)