FIGURE 1.

(A) A sample Luigi workflow. The workflow can consist of both linear and parallel tasks. A task that depends on a previous one will not be executed until all the dependencies have been completed. Tasks without direct or indirect dependency on each other can be executed in parallel if resource permits. (B) ARGem workflow. The ARGem pipeline automatically processes the raw sequences after a list of SRA accession numbers are submitted through a metadata spreadsheet. After preprocessing, the raw sequences go through two different branches: 1) short reads matching to generate normalized ARG counts, and 2) contig annotations against ARG and MGE reference databases. The results generated can then be passed to the integrated analysis and visualization tools. The default normalization methods built in the pipeline are 16S rRNA, TPM, and FPKM. 1) Blue rectangles indicate data and 2) red rounded rectangles indicate processing steps.