Novora · Literature-grounded workflow execution
Execute the plan, and check it against the literature
Novora runs the Nexora plan with full per-step provenance and reproducible-by-design output, then flags intermediate results that deviate from what comparable published studies reported, the same literature consensus that planned the workflow. Novora executes; it does not decide.
runs a pipeline and shows logs.
flags samples that look odd vs each other.
additionally flags results that deviate from published studies, in one end-to-end reproducible provenance chain.
The cited plan to execute.
Fixed for reproducible re-runs.
Supply processed data to execute stages for real (counts/feature-table/abundance + groups, or a line-list). Without it, stages are simulated and clearly flagged.
Understanding Novora
Novora executes the plan Nexora produced, captures every step as part of one end-to-end reproducible provenance chain, and flags intermediate results that deviate from what comparable published studies reported. It executes; it does not decide.
What Novora is
Novora is the execution arm of the closed loop. It takes the cited workflow plan from Nexora and the validated data from Veraxa, runs the analysis, and produces results ready for reporting, with a complete record of every input, parameter, version, output checksum and timestamp, so the run can be reproduced exactly.
The execution layer itself is deliberately standard. The value is not that Novora runs a pipeline, many tools do, it is that the run is one link in a single provenance chain from the literature (Lumen) through the design (Praxis), the data (Veraxa) and the plan (Nexora) to the report (Oris).
Why it's different
Runs your pipeline and shows logs and resource use.
Flags samples that look odd compared to your other samples.
Additionally flags intermediate results that deviate from what comparable published studies reported, using the same literature consensus that planned the workflow.
How an execution runs
- 1 · Translate the plan
The Nexora plan becomes an ordered set of executable steps with the parameter values already resolved from the literature.
- 2 · Execute step by step
Each stage runs on your data, producing intermediate outputs and quality-control metrics, with progress reported as it goes.
- 3 · Compare to the literature
Intermediate metrics are checked against what comparable published studies reported; deviations are flagged with the evidence behind the expected range.
- 4 · Record everything
Every step records its inputs, parameters, versions, output checksums and timestamps, a reproducible-by-design provenance record for reporting and audit.
Real, simulated, and honest
Where your input is processed, tabular data, Novora runs the analytical stage for real. Where a stage needs raw sequence or image data, it runs a clearly flagged simulation rather than pretending. Anomaly flags are equally honest: when comparable studies actually reported a metric, the flag is grounded in the literature with citations; when they did not, Novora falls back to a documented default threshold and says so, it never dresses a fixed threshold up as published evidence.
What you get
- Results ready for interpretation, differential tables, diversity outputs, trend summaries.
- Plain-language summary of whether results fall within the literature range.
- Intermediate-result anomaly flags, each labelled literature-grounded or default-threshold.
- A complete, reproducible provenance record, re-run an identical execution any time.