Fig. 1
From: HirBin: high-resolution identification of differentially abundant functions in metagenomes
An overview of the HirBin workflow. HirBin uses reference sequences as input to the annotation step. The reference sequences are annotated in terms of known functional domains such as TIGRFAM, PFAM or COG (supervised annotation). Each annotated domain is then clustered based on sequence similarity (unsupervised clustering) forming sub-bins. The sub-bin abundance is then calculated by mapping the reads in each sample to the sub-bin sequences and finally differentially abundant sub-bins between conditions are identified by statistical analysis