A computational research instrument
Sharper rankings.
Same interval lifetimes.
Change the origin of a persistence diagram. See which distances move, which ties split, and what the new ranking actually tells you.
Start with a synthetic example01 / Interactive laboratory
Follow the anchor
Distance along the path
Direct evaluations, with an endpoint envelope
Shading shows the inspected alternate’s endpoint bounds. Curves join directly evaluated samples; the highlighted points are evaluated at the current λ.
Birth–death coordinates
Reference and one shifted alternate
A translation preserves every interval lifetime.
A rank is a decision, not just a distance
Alleles share credit when their scores are tied.
| Alternate | Shift δ | Native | At λ 0.50 | Centered | Current top set |
|---|
A more decisive ranking does not establish better clinical discrimination. The synthetic selected label is illustrative, not a diagnosis.
02 / Use the instrument
Bring your own diagrams
Audit finite birth–death pairs and a declared shift for each alternate.
This tool starts from persistence diagrams. It does not infer diagrams from a contact map, select a biological anchor, or train a model.
- 1Supply the reference and alternate diagrams.
- 2Declare δ for each alternate. For mean centering, use the alternate-minus-reference map mean on the same edge support.
- 3Inspect the path and export distances, bounds and unresolved comparisons.
Browser limit: 20 points per diagram, 8 alternates, 2 MB per file. Empty diagrams and repeated points are supported. All coordinates must be finite with birth ≤ death. Use the Python toolkit for a larger audit.
Files are parsed locally. Their contents are never uploaded. The numerical allowance is practical floating-point tolerance, not a certified error bound.
Python toolkit & input formatPairs are [birth, death]. The optional selected field names an alternate; it does not influence distances.
03 / The genomic study
Decisiveness and discrimination
are different questions.
BMC Bioinformatics · submitted 29 October 2026 · Technical Check
The research audited GM12878 discovery and new-locus replication, then tested IMR-90 on the same 84 replication sites. The figures below are frozen study summaries; the laboratory above uses synthetic examples or your own inputs.
Loading the source-linked study summary…
04 / Educational companion
Understand the finding.
Keep its limits.
Ten short lessons connect the mathematics, the genomic results and their interpretation.
The printable editions cover the current rework, including all six supplementary files. Markdown editions and source coverage are also available.
Loading lessons…
Try it yourself
Check your understanding
Answer, check the explanation, and retry. This is a practice exercise, not a validated learner assessment.