1 Methods

1.1 Heatmap (unsupervised clustering)

Protein abundances were log2-transformed and z-score scaled across samples (per protein) to highlight relative patterns. Hierarchical clustering was applied to proteins using Euclidean distance and complete linkage.

1.2 PCA (unsupervised)

PCA was performed on the log2-transformed, z-score scaled abundance matrix (samples as rows, proteins as columns).

1.3 Volcano plot (exploratory)

Given the pilot sample size, volcano plots are used descriptively to visualize effect sizes (log2FC) versus p-values. Proteins are annotated based on large effect sizes (|log2FC| ≥ 1.5 folds) for interpretation, independently of statistical significance.

2 Results

2.1 Hierarchical Clustering and Abundance Heatmap

In the Hierarchical Heatmap below:

  • Each row is an individual animal; each column is a protein
  • Colors represent relative abundance (z-score): red = higher than average; blue = lower.

2.2 PCA (PC1 vs PC2)

2.2.1 Biplot

The PCA Biplot summarizes the proteomic profile into two main axes (PC1 and PC2). Separation between groups suggests consistent proteomic differences (exploratory).

2.2.2 PCA Loadings

The PC1 and PC2 loadings plot show the top 10 protein contributing to the components.

Proteins with larger absolute loadings contribute most to separation along the PCs.

The sign (positive/negative) indicates direction and should be interpreted with the PCA scatter.

2.3 Volcano Plot

The Volcano Plot show:

  • X-axis: how much proteins differ between CRA and OA (log2 fold-change).
  • Y-axis: strength of evidence (p-value), shown to guide visualization only.
  • Blu points indicate protein in the top 10 PC1 loading
  • Annotated proteins have large effect sizes (|log2FC| ≥ 1.5).

 

A work by DANILO LOFARO

danilo.lofaro@unical.it