Final matrix summary
- Genes / rows
- Awaiting preview
- Samples
- Awaiting selection
- Identifier mode
- Awaiting inspection
- Value domain
- Expression values; NQC validates raw-count compatibility
GeneBean · MARSS · RNA-seq workflow
Governed RNA-seq analysis from raw matrix to reviewable downstream data
Zomniverse provides a modular RNA-seq workflow for bringing expression matrices into a governed analytical state before downstream analysis. GeneBean and MARSS separate file inspection, row identity, study-matrix preparation, normalization, quality control, and downstream handoff into visible stages rather than treating the entire process as one opaque step.
Workflow stages
A modular path through RNA-seq preparation
The public workflow description below focuses on the role of each stage. Detailed internal algorithms, private implementation notes, and unpublished scientific methods are intentionally not exposed here.
-
01
File Reader
Ingests tabular expression data and checks whether the uploaded file has a usable structure before it can enter the scientific workflow.
-
02
Row Namer
Establishes row identity so the expression matrix has a consistent, reviewable feature axis before later transformations occur.
-
03
Expression Matrix Canonicalization
Organizes the intended sample columns and prepares a governed study matrix with clear dataset boundaries for downstream processing.
-
04
Normalization & Quality Control
Prepares compatible RNA-seq count data for review, exposes quality evidence, and keeps the normalization/QC stage distinct from later biological inference.
-
05
Governed downstream analysis
Committed outputs can be handed to compatible downstream modules. Research-stage capabilities such as GB-Tox are kept separate from the preparation stages so analytical responsibilities remain explicit.
Design principles
Why split the workflow into governed stages?
Inspectability
Users can see where the dataset is in the workflow and what kind of transformation or review belongs to each stage.
Reproducible boundaries
Separate stages make it easier to preserve clear inputs, outputs, and handoff points instead of silently changing the analytical state.
Human review
AI-assisted guidance can support the researcher, while governed data state and downstream execution remain distinct responsibilities.
Scientific context
From workflow infrastructure to biological interpretation
Zomniverse also publishes public research and tutorial material that shows how RNA-seq data can be explored after preparation. The GSE78220 melanoma analysis covers normalization, sample relationships, differential expression, pathway enrichment, and GSVA-based pathway activity.
