About ten years ago, I wrote an essay titled "Empowering users: self-service metabolomics data analysis for everyone?" The question mark was deliberate. At the time, the idea that any researcher could analyze their own omics data – without a dedicated programmer or statistician – was an aspiration rather than a standard reality.
Over the past ten years, we built a series of web-based tools, one omics type at a time – MetaboAnalyst for metabolomics, MicrobiomeAnalyst for the microbiome, ExpressAnalyst for transcriptomics, ProteoAnalyst for proteomics, and miRNet, OmicsNet, and OmicsAnalyst for networks and integration. Together they are now used by researchers around the world, and each one put self-service analysis within reach of more people. But each also stood on its own, and real studies rarely stay inside a single tool or a single omics type.
It was never a shortage of methods – there are always more methods than your data needs. The harder part is choosing the ones that fit the data in front of you, running them in the right order, and reading what comes back. A study often moves from one omics layer to the next, and the analysis has to move with it. That is the part we set out to solve this year.
Our answer is workflows. Each one is a verified, end-to-end path from raw data to result, and the building block everything else is composed from.
The Workflow Library now holds more than 130 of them. 100 are single-tool workflows – a two-group or time-series differential expression, a dose-response trend analysis, functional enrichment, a biomarker ROC model, a microbiome diversity comparison – each taking one upload and running end to end. Another 32 are advanced pipelines that chain those steps together, including multi-omics integration across transcriptomics, proteomics, and metabolomics. These are the building blocks.
They are yours to arrange. Run one on its own, fork it and adjust the parameters, or tick several and run them together on the same data. Chain them across omics types to carry a study from raw counts to differential expression to enrichment to a network in a single pass. And when the defaults do not cover what you need, save your own. You do not have to be a programmer or a statistician to do any of this – that was the point from the start. This is the freedom part: your data, and your path through it.
Freedom on its own is exactly where automated analysis tends to go wrong. An AI agent will happily assemble a pipeline that looks right and is quietly wrong. What keeps this honest is that every building block is grounded. Each workflow is built from the same peer-reviewed tools researchers have relied on for years – MetaboAnalyst, MicrobiomeAnalyst, ExpressAnalyst, and the rest – tested, traceable, and the same ones we teach and publish. Nothing in the library is a black box. For advanced users, you can ask your AI agent to extend the workflows to better suit your preference, or run a parallel method.
There is also a map. The workflows follow the framework in Omics Data Science, and the updated 2026 edition (2026v2) is written as the guide to them – the reasoning behind each analysis, when it applies, and how to read what it returns. The library gives you the pieces; the book tells you why they fit together. That is what makes this dependable rather than magical: freedom to compose, grounded in methods and a text you can stand behind.
Three pieces turn that library into a result you can check:
You bring your data, the agent composes the workflows that fit, and what comes back is a dashboard, a live report, slides, and publication-ready figures you can refine. About 20% of those analyses go past what the web interface can show, so they arrive in the report and slides. The interactive steps stay open to check by hand, and are yours to take over at any point.
And when there is no clear guidance on which parameters to use, the workflow does not stall. It runs across a sensible range of settings and returns one aggregated report on the landscape – slightly longer to compute, but automated, so it saves you significant hands-on time.
The work is far from finished. What comes next is about letting people shape the platform, not only use it:
From that first question about self-service analysis to OmicsVerse today, the aim has stayed the same: a dependable path from raw data to insight – free to compose, grounded at every step, and run on your own terms.
In the end, the software was never really the point. The point is the time it gives back – hours that once went into wrangling data can go back to the biology and the questions that made the work interesting to begin with. There is a long way still to go, and we would be glad to have you help shape where it goes next.
See it for yourself
Enter the platform and compose your first workflow, or read the book written to guide it.
Enter OmicsVerse Read the Book