Schematic of the MIST package. (A) Species-level typing. MIST maps the mNGS reads against the pan-genomes of each species and counts the species-specific reads for estimation of species abundance. (B) Strain-level typing. MIST starts by mapping the mNGS reads against all reference genomes. The mismatch information extracted from the SAM files is then converted to the probability of each read derived from each reference genome. This reads-based probability matrix is compressed into a cluster-based matrix according to the clustering results of the reference genomes. MLE is then imposed upon the matrix to infer the most likely abundances of each cluster in the mNGS dataset. Next, read ambiguity is used to quantify the abundance uncertainty along with the similarity of the sample’s components to reference genomes. Finally, all results are integrated and presented in a hierarchical manner. The module ‘Species’ is highlighted by the red box, the module ‘Index’ and ‘Cluster’ by the yellow box, and the module ‘Strain’ by the purple box.