LV R2X(%) R2X(cum)(%) R2Y(%) R2Y(cum)(%) Q2Y(%) Q2Y(cum)(%) 1 19.32 19.32 93.58 93.58 71.78 71.78 2 11.68 31.01 4.58 98.43 -8.26 69.45 3 14.86 45.86 1.19 99.62 23.94 76.77 R2X, R2Y: fraction of the variance of descriptor matrix (X) and class response (Y) explained by each latent variable (LV) in %, Q2Y: fraction of the variance predicted (cross-validated) in %; R2X(cum), R2Y(cum): cumulative explained variation and Q2Y(cum) predicted variation in %.

Table 1: Summary statistics of PLS-DA model for the SMA vs CN comparison. A three component model utilizes 45.86% of X for modelling 99.62% and predicting 76.77% of the response variation.