I am using MClust to cluster flow cytometry data (initial clustering meant for cleanup of the data) based on all the morphological parameters plus the UV channel (representing dead cells).
Specifying the noise parameter as those events (usually debris) that have very low FSC-A and SSC-A (cluster 2 in Figure 1) 
, I obtain an extra cluster (marked with "0", and since no cluster "0" is ever present without specifying the noise parameter, I suppose this should be actually representing the noise). Nonetheless, this cluster (Figure 2) does not appear to have anything to do with the original definition of noise, but instead seems to represent scattered events that do not fit in the remaining clusters (in this case I tried G=1:20 with modelName = "EEE". 
. I am wondering if it is just the fact that I am bending/misusing the concept of noise and if there is a way to include a noise parameter without specifying an initial vector of events to be interpreted as noise.
Thank you in advance