Figure 3.

Numbers of peak clusters occurring in each dataset. Peak clustering for spectra in the pooled samples (Set A) was performed independently of clustering for the individual spectra. 97 peak clusters were common to both sets.