Merge pull request #14 from RECETOX/issue6

Updated boxplot

analysis/Python_scripts/plotting.py

@@ -1,6 +1,8 @@
 import pandas as pd
 from matplotlib import pyplot as plt
+import seaborn as sns
 import plotly.graph_objs as go
+import numpy as np
 
 def make_boxplot(grouped_df: pd.DataFrame, colname: str, legend: str):
     #Create the plot with a width of 10 inches
@@ -79,4 +81,66 @@ def plot_histogram(x, xaxis_title='', title=''):
     fig = go.Figure(data=[trace1], layout=layout)
 
     # Display the plot
-    fig.show()
+    fig.show()
+
+def create_plot(df, path):
+    sns.set_style(style='white')
+    plt.figure(figsize=(17, 5))
+
+    # Set the color palette
+    colors = ['yellow', 'deepskyblue']
+    sns.set_palette(sns.color_palette(colors))
+
+    ax = sns.boxplot(x="true_names", y="value", hue="Number", 
+                    data=df, hue_order=['CosineHungarian_0.01_0.0_1.0_matches',np.nan],
+                    medianprops={'color': 'darkgreen', 'linewidth': 4.0},
+                    flierprops={'marker': 'o', 'markersize': 10, 'markerfacecolor': 'none'})  # RUN PLOT   
+    ax2 = ax.twinx()
+
+    sns.boxplot(ax=ax2,x='true_names', y='value', hue='Number',
+                data=df, hue_order=[np.nan, 'CosineHungarian_0.01_0.0_1.0_scores'], 
+                medianprops={'color': 'b', 'linewidth': 4.0}, 
+                flierprops={'marker': 'o', 'markersize': 10, 'markerfacecolor': 'none'})
+
+    ax.legend_.remove()
+    ax.set_ylim([0, 5])  # Set y-axis limits
+    ax.yaxis.set_major_locator(plt.MultipleLocator(1))  # Set major tick marks
+    ax.set_ylabel('Match values')  # Set y-axis label
+    ax.yaxis.label.set_size(20)  # Set font size of y-axis label
+    ax.set_xlabel('Chemical composition', fontsize=20)  # Set x-axis label and font size
+    ax.tick_params(axis='x', labelsize=13)  # Set font size of x-axis tick labels
+    ax.tick_params(axis='y', labelsize=13)  # Set font size of y-axis tick labels
+    ax.yaxis.labelpad = 10
+    ax.xaxis.labelpad = 10
+
+    # Create a count for each x-axis label
+    count_data = df['true_names'].value_counts().reset_index()
+    count_data.columns = ['true_names', 'count']
+    count_data = count_data.sort_values(by=['true_names'])
+    count_data['count'] = count_data['count'] // 2
+
+    # Remove the original x-axis labels
+    ax.set_xticklabels([])
+
+    # Add the count labels to the x-axis
+    ax.set_xticks(np.arange(len(count_data)))
+    ax.set_xticklabels(count_data['true_names'] + ' (' + count_data['count'].astype(str) + ')', rotation=45, ha='right')
+
+    ax2.legend_.remove()
+    ax2.set_ylim([0, 1])  # Set y-axis limits
+    ax2.yaxis.set_major_locator(plt.MultipleLocator(0.2))  # Set major tick marks
+    ax2.set_ylabel('Score values')  # Set y-axis label
+    ax2.yaxis.label.set_size(20)  # Set font size of y-axis label
+    ax2.tick_params(axis='y', labelsize=12)  # Set font size of y-axis tick labels
+    ax2.yaxis.labelpad = 10
+
+    # Change the legend labels
+    handles, labels = ax.get_legend_handles_labels()
+    labels[0] = 'Matches'
+    labels[1] = 'Scores'
+    ax.legend(handles, labels, loc='upper right', fontsize=14)
+
+    plt.savefig(path, dpi=300, bbox_inches='tight')  # SAVE PLOT AS PNG FILE
+    plt.show()
+    plt.clf()
+    plt.close()

analysis/Python_scripts/utils.py

@@ -1,5 +1,7 @@
 import pandas as pd
+import numpy as np
 from rdkit import Chem
+from itertools import combinations
 
 def is_spectrum_for_compound(compund_name, spectrum_name):
     options = [compund_name + x for x in ["", "_isomer1", "_isomer2", " isomer 1", " isomer 2"]]
@@ -45,11 +47,55 @@ def append_classes(df, left_on):
         "S": [has_atom(m, 'S') for m in molecules],
         "P": [has_atom(m, 'P') for m in molecules],
         "Si": [has_atom(m, 'Si') for m in molecules],
-#        "C,O,N,H": [has_organic_atoms(m) for m in molecules],
+        "C,O,N,H": [has_organic_atoms(m) for m in molecules],
+        "N": [has_atom(m, 'N') for m in molecules],
     })
     merged_df = pd.merge(df, class_names, left_on=left_on, right_on='molname')
     return merged_df
 
 # Define a function to map the true columns to a list of names
 def get_true_names(row, df):
-    return [col for col in df.columns[11:17] if row[col]]
+    return [col for col in df.columns[11:18] if row[col]]
+
+# Function to split values with commas and create new rows
+def split_and_add_rows(df, column_name, split_by):
+    df_copy = df.copy()
+    df_copy[column_name] = df_copy[column_name].str.split(split_by)
+    df_copy = df_copy.explode(column_name).reset_index(drop=True)
+    return df_copy
+
+def generate_combinations(df, column_name):
+    new_rows = []
+    for index, row in df.iterrows():
+        values = row[column_name].split(', ')
+        all_combos = []
+
+        for r in range(2, len(values) + 1):
+            for combo in combinations(values, r):
+                all_combos.append(', '.join(combo))
+
+        for combo in all_combos:
+            new_row = row.copy()
+            new_row[column_name] = combo
+            new_rows.append(new_row)
+
+    return pd.DataFrame(new_rows).reset_index(drop=True)
+
+def preprocess_data(merged_top5_same):
+    # Concatenate the DataFrames in df1_list and add a 'value' column with the value 'matches'.
+    df1 = merged_top5_same[['query', 'reference', 'true_names', 'CosineHungarian_0.01_0.0_1.0_matches']].copy()
+
+    # Concatenate the DataFrames in df2_list and add a 'value' column with the value 'scores'.
+    df2 = merged_top5_same[['query', 'reference', 'true_names', 'CosineHungarian_0.01_0.0_1.0_scores']].copy()
+
+    # Concatenate df1 and df2 into a single DataFrame.
+    df_cat = pd.concat([df1, df2])
+
+    mdf = pd.melt(df_cat, id_vars=['query', 'reference', 'true_names'], var_name=['Number'])      # MELT
+
+    #cleaning data
+    mdf = split_and_add_rows(mdf, 'true_names', split_by=', C,O,N,H')
+    mdf['true_names'] = mdf['true_names'].replace('', np.nan)
+    mdf = mdf.dropna(subset=['value', 'true_names'])
+
+    return mdf