Add initial R project file for xMap biomarker analysis

2026-05-12 11:01:52 +02:00
commit 7b124dbe07
29 changed files with 15190 additions and 0 deletions
@@ -0,0 +1,110 @@
+import nbformat as nbf
+
+nb = nbf.v4.new_notebook()
+
+text1 = """\
+# Part 1: Exploratory Data Analysis
+
+In this notebook, we explore the preprocessed datasets generated from the autoimmunity profiling study on Exfoliative Glaucoma (XFG).
+
+We will focus primarily on `slice_1_significant_7.csv`, which contains the 7 protein fragments found to be significantly associated with the condition.
+"""
+
+code1 = """\
+import pandas as pd
+import numpy as np
+import matplotlib.pyplot as plt
+import seaborn as sns
+
+# Load the slices
+slice_1 = pd.read_csv('slice_1_significant_7.csv')
+slice_2 = pd.read_csv('slice_2_sig7_plus_33_random.csv')
+slice_3 = pd.read_csv('slice_3_all_proteins.csv')
+
+print("Slice 1 (Significant 7) shape:", slice_1.shape)
+print("Slice 2 (Sig 7 + 33 random) shape:", slice_2.shape)
+print("Slice 3 (All proteins) shape:", slice_3.shape)
+
+slice_1.head()
+"""
+
+text2 = """\
+### Class Distribution
+
+Let's check the distribution of our target variable `group`, where `1` represents Exfoliative Glaucoma (XFG) and `0` represents Healthy Controls.
+"""
+
+code2 = """\
+plt.figure(figsize=(6, 4))
+sns.countplot(x='group', data=slice_1, hue='group', palette='Set2', legend=False)
+plt.title('Distribution of Target Variable (Group)')
+plt.xlabel('Group (0 = Healthy, 1 = XFG)')
+plt.ylabel('Count')
+plt.show()
+
+print(slice_1['group'].value_counts())
+"""
+
+text3 = """\
+### Feature Distributions
+
+We will examine the distributions of the 7 significant protein fragments across the two groups. This helps us visualize how well individual features might separate the classes.
+"""
+
+code3 = """\
+# Extract the feature columns (excluding identifiers and target)
+significant_cols = [c for c in slice_1.columns if c not in ['Internal LIMS ID', 'Original ID', 'group']]
+
+fig, axes = plt.subplots(nrows=2, ncols=4, figsize=(18, 10))
+axes = axes.flatten()
+
+for i, col in enumerate(significant_cols):
+    sns.histplot(data=slice_1, x=col, hue='group', kde=True, ax=axes[i], palette='Set2')
+    axes[i].set_title(col)
+
+# Remove the empty subplot
+fig.delaxes(axes[7])
+
+plt.tight_layout()
+plt.show()
+"""
+
+text4 = """\
+### Correlation Analysis
+
+Let's look at the correlation matrix to see if these 7 significant protein fragments are highly correlated with each other. If they are highly correlated, they might carry redundant information for our Decision Tree.
+"""
+
+code4 = """\
+plt.figure(figsize=(8, 6))
+sns.heatmap(slice_1[significant_cols].corr(), annot=True, cmap='coolwarm', fmt=".2f", vmin=-1, vmax=1)
+plt.title('Correlation Matrix of Significant Features')
+plt.show()
+"""
+
+text5 = """\
+### Motivating the Data for Classification
+
+This dataset is highly suitable for a binary classification task for the following reasons:
+1. **Clear Target Variable:** We have a well-defined, discrete target variable (`group`), which represents the presence or absence of Exfoliative Glaucoma (1 vs 0).
+2. **Numeric Features:** The autoantibody reactivity levels (median fluorescent intensities transformed via Box-Cox) act as continuous numeric features.
+3. **Biological Relevance:** The selected features (the 'Significant 7') have demonstrated statistical significance in separating the groups based on a Moderated t-test, providing a solid biological foundation that they contain predictive signal.
+4. **Iterative Model Building:** The slices provided (7 features, 40 features, and all features) will allow us to experiment with varying degrees of dimensionality and find the optimal representation to avoid underfitting or overfitting our Decision Tree and MLP classifiers.
+"""
+
+nb['cells'] = [
+    nbf.v4.new_markdown_cell(text1),
+    nbf.v4.new_code_cell(code1),
+    nbf.v4.new_markdown_cell(text2),
+    nbf.v4.new_code_cell(code2),
+    nbf.v4.new_markdown_cell(text3),
+    nbf.v4.new_code_cell(code3),
+    nbf.v4.new_markdown_cell(text4),
+    nbf.v4.new_code_cell(code4),
+    nbf.v4.new_markdown_cell(text5)
+]
+
+with open('EDA.ipynb', 'w') as f:
+    nbf.write(nb, f)
+
+print("EDA.ipynb created successfully.")