DNA-LLM/viral_complexity

1import pandas as pd2import numpy as np3from datasets import load_dataset4import matplotlib.pyplot as plt5from scipy.interpolate import interp1d6from shiny import render7from shiny.express import input, output, ui8# from utils import (9#     generate_2d_sequence,10#     plot_seq_full_label11# )12import os13import matplotlib as mpl14import seaborn as sns15mpl.rcParams.update(mpl.rcParamsDefault)16 17ds = load_dataset('Hack90/virus_tiny', keep_in_memory = True, cache_dir = None)18df_virus = pd.DataFrame(ds['train'])19 20def shannon_entropy(seq):21  seq=re.sub("[^ATCG]","",seq)22  seq = seq.replace('A', 'T')23  seq = seq.replace('G', 'C')24  p = seq.count('T') / len(seq)25  e = 8.69 - 8.3126  c_h = ((-p * math.log(p)) - (1-p)* math.log(1-p)) * math.log((1-p)/p)27  c_h = c_h /e28  seq=seq.replace('T', '5 ')29  seq=seq.replace('C', '4 ')30  seq = np.array(seq.split()).astype(int)31  shann = -sum((p*math.log(p), ((1-p)*math.log(1-p))))32  shann = shann/233  return c_h , shann34 35 36ui.page_opts(fillable=True)37 38with ui.navset_card_tab(id="tab"):39    with ui.nav_panel("Species View"):40        ui.panel_title("What is the distribution of complexity across viral species?")41        with ui.card():42            ui.input_slider("sample", "samples", 0, len(df_virus), 40)43                44        def plot_loss_rates(df,samples):45            complexity = []46            for k in range(len(df.iloc[:samples])):47              complexity.append(shannon_entropy(df['sequence'].iloc[k]))48            49            df_nana = pd.DataFrame(complexity)50            df_nana['x'] = df_nana[1] * 251            df_nana['y'] = df_nana[0]52 53 54            # fig, ax = plt.subplots()55 56 57            # Create a figure and axis58            fig, ax = plt.subplots()59            60            # Create the scatter plot61            scatter = ax.scatter(df_nana['x'], df_nana['y'], s=0.5)62            63            # Add a colorbar64            cbar = fig.colorbar(scatter, ax=ax)65            cbar.set_label('Label')66            67            # Set labels and title68            # ax.set_xlabel('X')69            # ax.set_ylabel('Y')70            # ax.set_title(f"Loss ra")71            # ax.set_xlabel("Training steps")72            # ax.set_ylabel("Loss rate")73            return fig74 75        @render.plot()76        def plot_context_size_scaling():77            fig = plot_loss_rates(df_virus,input.sample() )78            if fig:79                return fig80    # with ui.nav_panel("Histone Modification"):81    #     ui.panel_title("Is there a pattern to histone modification?")82    #     with ui.layout_columns():83    #         with ui.card():84    #             ui.input_slider("sample_histone", "sample", 0, df_histone_len, 40)85        86        87    #     def plot_histone(df,sample):88    #         y_values = generate_2d_sequence(df['seq'].iloc[sample])[0]89    #         x_values = generate_2d_sequence(df['seq'].iloc[sample])[1]90            91    #         integers = str((np.argwhere(df['labels'][sample] == np.amax(df['labels'][sample]))).flatten().tolist())92    #         # Create a DataFrame with the x values, y values, and integers93    #         data = {'x': x_values, 'y': y_values, 'color': integers}94 95    #         fig, ax = plt.subplots()96 97    #         sns.scatterplot(x='x', y='y', hue='color', data=data, palette='viridis', ax=ax)98    #         ax.legend()99    #         # ax.set_title(f"Loss ra")100    #         # ax.set_xlabel("Training steps")101    #         # ax.set_ylabel("Loss rate")102    #         return fig      103    #     @render.plot()104    #     def plot_histones_two():105    #         fig = plot_histone(df_histone,input.sample_histone() )106    #         if fig:107    #             return fig108    # with ui.nav_panel("Enhancer Annontations"):109    #     ui.panel_title("Is there a pattern to enhancer annotations?")110    #     with ui.layout_columns():111    #         with ui.card():112    #             ui.input_slider("sample_enhancer", "sample", 0, df_enhancer_annotation_len, 40)113    #     @render.plot()114    #     def plot_enhancer():115    #         fig = plot_loss_rates(df_enhancer_annotation,input.sample_enhancer() , True)116    #         if fig:117    #             return fig            118        119