Generally, MIMOSA2 integrates some form of microbiome dataset with a paired dataset of measurements of identified metabolites. It can analyze various different combinations of types of microbiome and metabolite data.
Specifically, microbiome data can be provided in the form of either a taxa-based table of 16S rRNA microbiome data processed in one of several ways, or a function-based table of KEGG Ortholog abundances annotated from metagenomic sequencing data. Metabolite data can be generated from any metabolomics platform (e.g. LC-MS, GC-MS, NMR…), but should consist of processed measurements of identified metabolites (recommended to have KEGG compound IDs).
Each data file must be formatted as shown in the examples linked below. Input data files should be tab-delimited or comma-delimited tables with a row for each feature and a column for each sample, with the first row being a header of names for each column. There should not be duplicate features or samples. If sample IDs differ between the microbiome and metabolome data files, MIMOSA will only analyze samples that are present in both. MIMOSA currently normalizes microbiome data to relative abundances by default. Limited options for metabolomics data normalization are provided.
When using the web application, wait for the progress bar for all of your files to indicate “Upload complete” before starting the analysis. If you find they are taking a long time to upload, this is likely due to excessive load on the server - please be patient or try again later.
Microbiome composition data can be provided in 3 different formats:
1) Sequence variants (ASVs): 16S rRNA data processed by denoising tools (i.e. qiime2/DADA2/Deblur). In this case, the first column must contain the DNA sequences themselves. If you provide ASV data, you can also control how strictly your sequence variants are mapped to the reference database of 16S rRNA sequences linked to metabolic reconstructions (see Settings).
2) Greengenes 13_5 or 13_8 OTUs: 16S rRNA data assigned to closed-reference Greengenes OTUs (e.g. by QIIME1 or vsearch).
3) SILVA 132 OTUs: 16S rRNA data assigned to closed-reference SILVA OTUs. Note: SILVA OTUs are not currently compatible with the KEGG metabolic network database option.
4) Metagenome: Total KO abundances: A table of function abundances, as produced from any metagenomic functional annotation pipeline. Note: The name of the first column must be “KO”. Example total KO table
5) Metagenome: Taxon-stratified KO abundances (HUMAnN or PICRUSt/PICRUSt2): A table of taxon-specific function abundances, which can be produced from HUMAnN, PICRUSt/PICRUSt2, or similar programs (e.g. kraken). This must be formatted either in the style
of the “metagenome contribution table” produced by PICRUSt version 1 or 2 (see the PICRUSt2 documentation), or
in the format of the stratified abundance table produced by HuMAnN. MIMOSA2 will detect which format is used and process your table accordingly. For the PICRUSt/PICRUSt2 format, the columns can be in any order, but they must include columns named “Gene”, “OTU”, “Sample”, and “CountContributedByOTU”. If you have a metagenome table from HuMAnN with abundances of UniRef gene families or EC numbers, you can map it to KOs using the humann_regroup_table utility function in HuMAnN.
Example HuMAnN format
Example PICRUSt2 format
MIMOSA only analyzes metabolite features that can be assigned a KEGG compound identification. Compounds can be either be provided directly as KEGG IDs, or as metabolite names, which are mapped to KEGG IDs using MetaboAnalystR. However, this process is generally improved by a manual curation step, so we encourage you to run the MetaboAnalystR utility yourself at their website. Then you can manually curate the annotations and use the resulting table for MIMOSA2 analysis.
Example metabolites (KEGG IDs) (Note: the name of the first column must be “KEGG”)
Example metabolites (metabolite names)
A single table can be provided to specify multiple types of modifications to the model. Modifications include adding and removing KOs and/or reactions, and these modifications can apply to all community members or to specific taxa. Formatting of this modification file is illustrated in several examples below: