J-EXPRESS / ROADMAP

What works today.
What comes next.

Our direction for J-Express: approachable scientific analysis, transparent AI collaboration and more kinds of data in one exploratory workspace.

Reviewed 4 November 2026. This is a curated summary of our development plan. Future priorities may change as researchers bring real datasets and new questions; no release dates are committed.

Open J-Express ↗

01 / DELIVERED

Available foundations.

  • Interactive matrix exploration: PCA, correspondence analysis, clustering, heatmaps, SOM and measurement profiles, with linked selections and groups.
  • Biological workflows: bulk and single-cell RNA-seq exploration, QC and the Airway and PBMC 3k demos; existing J-Express projects and numeric tables.
  • Embedding exploration: PCA/UMAP, cosine neighbors, semantic search and concept directions, with text demonstrations.
  • AI connectivity: authorized MCP dataset access and supported analysis operations, backed by documented scientific workflow procedures.
  • NGS result interoperability: upstream QC and RNA-seq result import, including a verified nf-core/rnaseq example. Alignment and quantification remain upstream.
  • Pathway analysis: initial GSEA/GSVA delivery through optional local-runtime adapters, including gene-set review and derived score datasets.

Some statistical methods require an explicitly connected local R/Python environment and separately installed packages. Availability does not imply suitability for every experimental design or independent scientific validation. Read the scientific guidance ↗

02 / ONGOING WORK

Strengthen the workflows.

  • Browser-first computation: extend browser-native pathway analysis and selected single-cell correction methods, with numerical comparisons to established implementations.
  • Reliable scientific orchestration: complete local-runtime end-user acceptance, broaden real-data validation and keep AI workflows aligned with executable tools.
  • Usability and integration: improve onboarding, method guidance, API coverage and reproducibility across the analysis flow.

Implementation, production deployment and validation are separate milestones. Follow-up work is described here as ongoing until those checks are complete.

03 / PLANNED DATA EXPANSION

More data.
The same curiosity.

These are proposed extensions, rather than claims of dedicated support today. A compatible numeric table is a useful start; full support also needs appropriate QC, transformations, metadata and scientific guidance.

  1. Proteomics, metabolomics and lipidomics: our first proposed new data-family delivery. Start with processed abundance tables, missing-value handling, measurement scales and sample/batch metadata.
  2. Embeddings from images, audio, DNA and proteins: import existing vectors with model provenance and compatibility checks.
  3. Spatial omics: link expression patterns to tissue or cell coordinates and spatial selections.
  4. Epigenomics: methylation, ATAC-seq and ChIP-seq result matrices with genomic-region metadata and assay-aware transformations.
  5. Time series and sensors: timestamps, measurement units, uneven sampling and linked temporal exploration.

Further candidates include flow and mass cytometry, microbiome data, clinical and phenotypic cohorts, multi-omics, genetic variation/GWAS and networks or distance matrices.

We plan to build on processed results from established tools, preserving their provenance and scientific meaning.

04 / FUTURE DIRECTIONS

Scale and explain.

  • Larger datasets: profile and improve memory use, projections, clustering, heatmaps and interactive responsiveness.
  • Additional methods: evaluate HDBSCAN, NMF and WGCNA; consider trajectory and model-aware interpretability methods when concrete workflows justify them.
  • Broader interoperability: evaluate variant-result import and optional external pipeline orchestration after the existing import workflows are established.

New formats and methods will be scoped against real examples, scientific assumptions and practical value before implementation.

05 / HELP SET THE DIRECTION

Bring a dataset.
Bring a question.

Tell us which data you work with, where your workflow gets difficult and what you would like to explore. Representative datasets and domain expertise help us choose useful next deliveries.

Contact Resonalis ↗