Table 2.
Quality assessment metrics. The set of predicted bins and expected bins (gold standard) are, respectively, represented by 
and 
. Each predicted bin 
is assigned to its best mapping expected bin 
before computing purity, contamination and completeness. 
equals to the number of genes in the predicted bin 
and 
represents the set of expected bins that do not correspond to the best mapping of any predicted bin and are associated with a completeness of 0. NMI corresponds to the Mutual Information (MI) normalized by the maximum of the unconditional entropies of the expected and predicted bins 
and 
represents the definition of MI where the normalized number of shared genes between a given predicted bin 
and an expected bin 
is defined as 
; and the normalized number of genes in 
and in 
are 
and 
.
