Figure 5. Correlation matrices for PramilaL dataset (dual-channel), for each normalisation technique.

Enlarged areas, labelled (a), (b), (c), (d), (e) are provided for better visualisation. Behaviour similar to that for the Hasse dataset can be observed, with decreased amount of high correlation for XPN cross-platform normalisation, and with high correlations for PMOnly methods. However, differences are smaller than for the Hasse dataset, indicating that single-channel data is more sensitive to the normalisation approach taken.