Motivation & Vision
Empowering researchers through Training, Tools, and AI — our mission for over a decade.
Three Stages of Omics Analytics
Generating data is no longer the hard part — analysis is the bottleneck. And the way we analyze omics data has moved through three stages, each with its own limiting step.
Analytics 1.0, the expert era, meant writing your own code; trained programmers were the bottleneck. Analytics 2.0, the web era, replaced code with point-and-click tools that made the methods learnable by anyone — where our free platforms reached over a million researchers — but the rigid interface became the new limit: a menu can only offer what its buttons allow. Analytics 3.0, the agentic era, lets you direct AI-composed workflows in dialogue with your data, automating the analysis — and the bottleneck shifts again, to ideas, computing, and trust.
From writing code (1.0), to point-and-click tools (2.0), to directing AI-composed workflows (3.0) — the bottleneck moving from programmers, to the interface, to ideas and trust.
A Reality Check on AI
That third stage is exciting, but it deserves honesty about what today's AI can and cannot do. Current models are strong coding and thinking partners — they write and debug analysis code, suggest sensible methods, and explain what a parameter does or how to read a plot. Where they struggle is the heart of a real study: wrangling messy, inconsistent data, planning the many precise, sequential steps a question demands, and catching an early mistake before it propagates.
The hard part of omics analysis was never a shortage of methods — there are always more methods than the data needs. It is to identify and apply these methods in right order. That is exactly what AI has least of, and what a researcher needs most. The answer is to give it verified building blocks (i.e. analysis patterns and validated workflows) to compose, not a blank slate to improvise on.
AI is strong at
- Writing and debugging analysis code
- Suggesting a method or explaining a concept
- Explaining a parameter or how to read a plot
- Small conveniences — tidy a file, refine a figure, draft a report
AI struggles with
- Wrangling messy, heterogeneous data
- Planning the many precise steps a real question needs
- Recovering from failure and iterating to a correct result
- Running a full analysis reliably without supervision
OmicsVerse is our path to Analytics 3.0. It runs on a library of verified analysis workflows — the same peer-reviewed methods from our web tools, now composable. You can drive it two ways: manual mode, the classic point-and-click interface and still the best way to learn the methods; or workflow mode, which chains those verified steps into multi-step analyses the menus never offered — across omics types, batched over many datasets, end to end.
That composability works on its own, before any AI is involved. And when you do add AI, the workflows ground it: instead of improvising a pipeline from raw data, the assistant composes established, inspectable building blocks — so every result traces back to a tested method, and you remain the final judge. It runs local-first, on your own machine.
