As we know from the Bruker's instruction or CLSI, the recommended bacterial load for identification by MALDI-TOF MS is around 10^5 CFU. Moreover, it is also clear that species identification would be wrong when the bacterial load is low in Bruker's MALDI-TOF MS. On this basis, however, I am curious why we fail in getting reliable identification when the bacterial load is low. Is it the problem attributed only to low peak intensity or bad S/N ratio? What would the MS spectra look like? Does anyone has such the experience? Do we have opportunity to improve the identification under the condition by more advanced bioinformatics?