Step 1: Build the microbiome dataset. Start with real reference samples, choose the changes you want to simulate, and see what the analyses find. You can then use this dataset to generate survival outcomes.

See the big picture: how many samples and microbes your dataset contains, and which groups of microbes are most common.

Simulation summary


                        

Look at the numbers behind the plots. Rows are microbial features (OTUs), and columns are samples.

Display limits do not change the simulation or full downloads. Tables and CSVs use sample/otu labels; original IDs are preserved in the mapping and RDS.
Download original ID mapping

Ask three questions: how diverse is each sample, how different are samples, and which microbes change with your study variable?

Alpha diversity ?

Inverse Simpson is 1 / sum(p²). Chao1 estimates richness using singletons and doubletons. Use Rarefy to subsample each eligible sample to the same read depth, then recalculate this alpha-diversity plot and its test.
Download counts used for alpha diversity

Beta diversity ?

Diversity tests assume independent samples. Their p-values are unadjusted for confounders and multiple testing; associations do not establish causation.

Differential abundance: true versus detected signal ?

The true panel highlights only simulator-assigned OTU tips. The detected panel combines ANCOM-BC2 results from OTU, Phylum, Class, Order, Family, and Genus levels where taxonomy is available. Each blue node passed its own test; a detected parent does not make its descendants detected. All other nodes are gray. Both panels share the same layout.
The table lists true-signal OTUs and their OTU-level results. The higher-taxa column lists detected ancestors from separate rank-level tests. Bold adjusted p-values meet the significance threshold. The complete download contains all features at every tested level.
Download differential-abundance results Download ANCOM-BC2 output
Step 2: Add follow-up outcomes. Use your microbiome dataset to create event times and follow-up records. Explore whether microbial differences are linked to earlier or later events. You can also upload your own count data.

Get a first look at your simulated study: its microbes, observed events, and survival patterns over time.


                        
Node area represents mean relative abundance across nonzero-depth samples in this saved survival dataset. Higher-level nodes sum their descendant OTUs. Labels show phyla; gray tips are OTUs.

See each person's follow-up time, whether an event was observed, and their microbiome counts. Download the complete study data here.

Outcomes and counts for the selected number of samples and taxa. Downloads include all data. Original IDs are preserved in the mapping and RDS.

Explore whether people with different microbiomes have different survival patterns, and which microbes are linked to earlier or later events.

Alpha diversity and survival ?

Download alpha diversity + survival

Taxon associations with survival ?

A separate Cox PH model relates each OTU's CLR abundance to observed time and event status. P-values are from likelihood-ratio tests, with BH adjustment across all input OTUs. HR > 1 indicates higher hazard per one-unit increase in CLR abundance. These are unadjusted associations.
The plot shows estimable OTUs. Blue points meet adjusted p < 0.05. The table shows up to 50 OTUs ordered by adjusted p-value; the CSV contains all OTUs. Selected timing/occurrence taxa describe the simulation and are not necessarily the same as marginal Cox associations.
Download all taxon–survival results

Beta diversity and survival ?

Placeholder: beta-diversity survival analysis will be added here.
Step 3: Explore study size. Repeat the study using your saved simulation settings to see how detection changes as you add more samples. Each repetition creates a fresh dataset.

See how often an analysis finds a signal as the study gets larger. The red line marks an 80% target.

The red line marks power = 0.80. Vertical bars show ± one Monte Carlo standard error, not 95% confidence intervals. Each point and SE use only completed, valid repetitions at that sample size and update as results arrive. At least two valid repetitions are needed for SE; identical results give zero-length bars. Failed tests are excluded and counted separately. Five repetitions yield coarse estimates.

Review your results, the settings used, and what happened in each repetition—all in one report. Download a complete report to keep or share.

Summary results

Valid counts usable results; Failed counts tests that could not produce a usable result. Power uses valid repetitions only. Planned is the number requested per sample size. Mean_FDR summarizes false discoveries among detected OTUs; Mean_event_proportion is the observed-event fraction in survival analysis.

Simulation settings


                        

Repetition details