Microbiome Explorer Tutorial

Data Preparation

File upload

Microbiome Explorer accepts several different file formats containing metagenome sequencing results which will be stored at a MRexperiment internally. If an MRexperiment is already available, the user can simply upload it as an rdata or RDS object (the application will attempt to guess the correct file format based on the file extension). Otherwise, Biological Observation Matrix (BIOM) formattes files produced by any program including qiime2 or mothur can also be provided. Or, a simple counts matrix in csv or tsv form can be uploaded. A counts matrix should have no row names and store the information on the features/OTUs in its first columns. All other columns names should correspond to sample names. If the data does not contain phenotype information or in case the user wants to adjust the phenotype information included with an MRexperiment or biom file, they can upload an additional phenotype file and link it with the counts data. The required format is such that each sample is a row with the names of the rows in the phenotype data corresponding to the names of the columns in the count data. Appending several phenotype files by subsequent uploads is possible. If no phenotype file is given, the names of the columns of the count data are used as the phenotype data. Finally, if not already included with a given MRexperiment, a feature data file must be provided if aggregation to a particular phylogenetic level is desired. Here, each unique feature is a row which must correspond to the ones in the counts data feature column and each column is a taxonomy level. All three data entities (counts, phenotypes and features) will be stored together internally as an MRexperiment. This can be downloaded (if desired) via the “Get Data” button on the Data input page of the Microbiome Explorer.

Data Filtering

Once the data has been uploaded, several QC plots give an overview of the number of features and the number of reads available in each of the samples. Here, the user can filter sample via several sliders requiring e.g. a minimum number of features present in a sample or a minimum number of samples that a feature needs to be present in. Any changes in slider settings are immediately shown in the QC plot to allow the user to see the effect such filtering would have. In addition to threshold based filtering, the user can also decide to subset by specific phenotypes. Here, any levels that are not of interest should be selected (these will not be reflected visually in the QC plot). Once the user presses the “filter” button, the data structure is subsetted accordingly. In order to get back to the original dataset, a “reset” button is available.

In addition to the basic QC plots showing distriution of number of reads and number of features, a sample based barplot gives an overview of the number of base features (e.g. OTUs) available for each sample. This can be arranged based on frequency or phenotype levels as well as colored by phenotypes via the plot options.

All of the plots will be included in the reports if the user presses the “report” button on this page. Otherwise, filtering will be included (to reflect the data that was analyzed), but the QC plots will be omitted.

Phenotype modifications

The phenotype tab shows the available phenotype associated with the feature count set in an interactive table. The main purpose of this table is informational, but if desired the user can create new phenotype columns by comibining existing ones. This enables the creation of new data to center the analysis around: the values of the two columns are pasted together creating an interaction value. The user can furthermore select columns to show in the table and remove those irrelavant for the analysis. Finally, if desired, the data type of each column can be modified by opening the adjust datatype box via clicking on the “+” icon. This is usually not required, but might be of interest for specific analysis requests. Any modifications made are only stored in the internal MRexperiment for the analysis after pressing the “save” button. All saved modifications are automatically included in the reports.