Algorithm 1 (A finite mixture and latent Dirichlet model for metagenomics).

Require the hyperparameter $\overrightarrow{\alpha }$ , the total number of reads $N$ , the species set $G$ , the read-composition distributions $\mathsf{\Phi }$ Ensure the read dataset R sample the species mixture parameters $\overrightarrow{\theta }\sim Dir(\overrightarrow{\alpha })$ for a metagenome repeat 1. sample species $g_i\sim Mult(\overrightarrow{\theta })$ 2. sample read $r_k\sim Mult({\overrightarrow{\phi }}_{g_i})$ until the total number of metagenomic reads $N$ is reached return the read dataset $\text{{}{r}_1,r_2,\cdots ,r_N\text{}}\triangleq R$