OverviewEasyMap is an interactive web tool for evaluating and comparing associations of clinical variables and microbiome composition. This interactive online tool allows:
(1) Running multiple multivariate linear regression models, on the same features and metadata.
(2) Visualizing the associations between microbial features and clinical metadata found in each model.
(3) Comparing across the various models to identify the critical metadata variables and select the optimal model.
Step 1: Input data uploadPlease upload your input data here. The file should be a separator based file including all relevant data: clinical metadata variables and taxonomic features abundance for each sample (csv, tsv etc.)
Once you upload your file, all the identified columns will appear and 'Submit' will take you to the next step of analysis. If there is any problem with parsing the file, the error will be presented here. Once you fix the problem, try to reload the file.
Run an exampleClicking on the ‘example’ button will load a case study data as an example. In this case study, we considered six clinical variables that were collected in our cohort, and are also known to have an impact on gut microbiome composition: mode of delivery (vaginal or C-section), age (at time of visit), ethnicity, use of probiotics in the first year of life, infant diet at each time point (breastfed, formula-fed, mixed), and infante allergy status (case/control). In this study we searched for microbial features that are associated with the allergy status, taking into account all other clinical variables. For more details regarding this case study please check our paper .
Clicking the 'Load example data' will load the example data to the app which will then enable you to continue to analyse the example data by clicking the 'Submit' button above . You also can review the exact format by downloading the example data file itself by clicking the 'Download example data' button.
Step 2: Variables type definitionHere, define the type of all clinical metadata variables. In each one of the categories below choose the suitable variable. Please read the definitions beside each category and carefully choose the variables.
Step 4: A birds-eye view of all significant resultsHere, you can compare the models using a high-level comparison of all microbial associations that were found to be significant in at least one model. Heatmap color displays the significance, and the default FDR q-value threshold is set to be 0.2 (only associations that pass this threshold appear in color). You can further filter the presented microbial features, using the drop-down menus on the left. In addition, you can select a different threshold, and choose which models to include in the heatmap.
Step 5a-b: Detailed view of selected associations and Alternating views of the raw data of selected associationsHere, you can toggle quickly between the bird’s eye view of all associations in the heatmap and zoom in on specific associations of interest. When you hover on a single cell in the heatmap, the cell is highlighted, and the relevant microbial feature together with the selected model, and associated clinical variable appear as text at the bottom of the panel. When you click on a certain cell in the heatmap, This panel is populated with a detailed plot showing the relative abundance (AST, if it was transformed) of the selected microbial feature by the selected clinical variable (this can be either a box plot for a categorical variable or a scatter plot for a continuous clinical variable). Note that if the relative abundance values (y-axis) are arc-sinus transformed then thus can exceed 1, and range in [0, 1.57079]. This plot also displays the q-values that are outputted by MaAsLin2 for all tested associations in this variable (using brackets comparing each value to the selected reference). Significance analysis appears for all possible comparisons between the reference and other values, with their respective q values, even for the non-significant comparisons.
Finally, to include additional metadata to the existing plot, you can facet the box plot and/or color the dots, by a specific variable. When the plot is faceted, the MaAsLin2 q-values are removed from the plot, and instead a t-test is performed, and p-values are presented. Finally, the user can color the dots based on the categorical variables of the model, and add labels to the dots, based on the random variables of the model.
All tests are between groups on the X axis (even if the user split and color samples by a specific variable). The Q value represents comparison between the full groups, except for cases where the facet option is selected and then a t-test is performed on each subset separately.
Q-values assigned as ‘q’ and coefficient values as ‘c’ on the plot.