Job 02 · Compare
Your cohort against a reference, with the differences in plain view.
Before agreeing on a comparison we review the compatibility of assay, references, population and processing. Applying the same downstream analysis does not remove differences at the source.
The result
Assess what can be compared and which differences prevent a conclusion.
A processed cohort, a declared comparison and an explicit list of what the comparison does not allow you to conclude.
Available The public 16S atlas can be consulted. Comparing a new cohort requires a compatibility assessment and an agreed method; it is not a self-service capability.
What is included
Compatibility before computation
We document which stages can be harmonised and which differences in laboratory, sequenced region or taxonomy remain. Whatever remains is not attributed to biology by default.
The limits, declared
We write down which differences in design, laboratory and population prevent a conclusion. That list travels with the result.
The next batch
The contract is written so that next year's cohort is processed the same way as this year's.
Snakemake · microbiome-counts-analysis 1.0.0 · Synthetic labels · OLS on CLR · BH · seed 1729
The proof, checkable now
90
features in the count table
The rows in counts.tsv contain synthetic features used to test the analysis and recover known signals.
- Dataset
- nexalel-review-challenge.1729
- Method version
- Snakemake · microbiome-counts-analysis 1.0.0
- Run identifier
- Frozen example · verifiable scientific fingerprint
- Artifact hash
- 34211cd306d23d5e22b36e26d4d0a5ae9f4714b2c118e7c294a4afb627e0fae6
- Seed
- 1729
The check runs in your browser and confirms file integrity. It does not replace scientific review or a fresh run.
What held up
A diversity signal holds under the documented depth check.
What did not hold up
Of the two significant contrasts, neither survived the sensitivity test. We publish that too.
We publish both. A method that only shows its successes cannot be evaluated.
Download the synthetic tables and test the analysis ↓ · 0.01 MB · No account, no form and no sales call.
What does your next study have to demonstrate?
We first agree on the question, the assay, the data rights and the deliverable. Then we decide whether working together makes sense.