Science
A simulated pipeline today, and a contract for real pipelines tomorrow.
Nexalel runs containerised bioinformatics pipelines with fixed versions: a microbiome workflow in R and Snakemake and a genomics evaluation on public cases. The same interface and the same artifact contract receive every new method.
Input
VCF files and tables with chromosome, position, reference and alternate columns. The structure is validated before processing.
Processing
Five stages with progress events: validation, normalization, annotation, prioritization and reporting.
Output
Manifest, summary, quality control, variant table, HTML report and provenance, with checksums and a contract version.
What it does not do
It does not compute disease risk, does not diagnose and does not replace expert review. All output is marked as simulated.
Two cohorts, two processing paths.
The downstream analysis shares a single pipeline version. The source reads, upstream processing and taxonomic references differ, and that difference limits direct comparison between cohorts.
Nexalel synthetic challenge
nexalel-review-challenge.1729
Data generated to verify the software, its sensitivity and its design warnings. It does not represent patients and does not constitute clinical validation.
Snakemake · microbiome-counts-analysis 1.0.0 · OLS on CLR · BH
Research · not for clinical use
Every finding needs context.
The framework connects observation, quality, source, uncertainty, permitted use, research question and report. Collaborative review per study is in development.
Observation
The public case presents computations on cohort data. The genomics and microbiome demos use synthetic fixtures. Each experience states its origin.
Inference
Annotations, associations or predictions that depend on references, models and assumptions. They are not direct measurements.
Interpretation
Conclusions limited to the available context and evidence. These demos offer no diagnosis or prescription.
Methods and background reading
These references explain the approach and its limits. They do not validate the demo's synthetic figures and imply no affiliation.
- International consensus on microbiome testing ↗
Clinical utility, standards and limits of interpretation.
- QIIME 2 · microbial marker gene analysis ↗
Methodological reference for amplicon analysis. The methods actually used are detailed in the case report.
- Classifier validation across cohorts ↗
Performance in one cohort does not guarantee generalization to others.
- Limits of co-occurrence networks ↗
A statistical association does not demonstrate a causal interaction.