Precision circRNA therapeutics
Find thecandidatesthat matter.
Back-splice junction evidence, host-gene association, and miRNA sponge activity — scored together into one defensible ranking, with the dataset, run, and decision still attached.
Admin-provisioned · Institutions & research labs
Rank 1 · Cohort A
hsa_circ_0000190
AKT3 · Mesenchymal
0.00
evidence score
- Host-gene associationAKT3
- miRNA sponge sitesmiR-21-5p · 4 sites
- Sequence conservation0.88
- Cohort recurrenceMesenchymal
Every score decomposes into the evidence behind it — nothing is a black box.
Anatomy of evidence
One record, fully dissected.
Every evidence record carries its grade, its independence key, and its source — so duplicate extractions can never masquerade as independent confirmation.
- candidate
- hsa_circ_0000190
- Canonical identity — never a transcript name.
- evidence grade
- B
- The weight this record carries, A through D.
- independence key
- LINEAGE-SEED-0
- Records sharing a key are one source lineage — not independent replication.
- context
- CTX-GBM-MES
- The biological context the claim is scoped to.
- mechanism family
- miRNA_ceRNA
- The competing mechanism this record supports.
- source
- PUB-SEED-001
- Where the extraction came from — always attached.
01 · Data
Data arrives in a governed workspace.
Sequencing runs and external sources land in a lab-scoped workspace. Inputs are schema-validated and samples linked on the way in — so every downstream result can name the data that produced it.
- DS-5012GBM Cohort A — Primary TumorsRNA-seq · Samples linked · schema validated48Ready
- DS-5027Recurrent GBM — Longitudinalpaired · Validating read pairs26Validating
- DS-5044IDH-mutant AstrocytomaRNA extract · Awaiting sample manifest12Staged
Every downstream result names the dataset and sample set it came from.
02 · Analysis
Compare runs, not screenshots.
Every analysis is scoped to a cohort and a run, with the pipeline version recorded alongside. Compare two runs to see which candidates moved and why — without reconstructing context from file names.
run_2041_v3
Cohort A · 48 samples
Showing Run 4 · current, run_2041_v3
- hsa_circ_0000190AKT30.94Unchanged
- hsa_circ_0001649SHANK10.89Unchanged
- hsa_circ_0072088ZFR0.86Rank up
- hsa_circ_0014717SMARCA50.83Rank down
- hsa_circ_0008362RTN40.77New this run
Scores shown on the platform's 0–1 scale. Switch runs to see which candidates moved.
03 · Candidates
A shortlist you can interrogate.
Open any candidate to its back-splice junction, host-gene locus, and miRNA sponge profile. Record the decision, and the reasoning stays attached to the evidence it was based on.
hsa_circ_0000190
chr1:243,663,021-243,777,047 · AKT3
Evidence decomposition · score 0.94
- Host-gene associationAKT3
- miRNA sponge sitesmiR-21-5p · 4 sites
- Sequence conservation0.88
- Cohort recurrenceMesenchymal
Interaction record
miRNA sponge profile
- miR-21-5p4 sites · -28.4 kcal/mol
- miR-155-5p2 sites · -22.1 kcal/mol
- miR-128-3p3 sites · -19.7 kcal/mol
RBP interactions
Pathways
04 · Validation
Move a candidate forward without losing context.
Divergent primers spanning the back-splice junction, RNase R controls, and siRNA knockdown assays — planned per candidate, then run in-house or handed to a connected provider.
hsa_circ_0000190 · handoff
- Prioritised
- Protocol attached
- Sent to provider
- Results returned
- hsa_circ_0000190FollowedAKT3qPCR + siRNA knockdownExternal providerIn Progress
- hsa_circ_0014717SMARCA5siRNA knockdownIn-houseIn Progress
- hsa_circ_0001649SHANK1qPCR validationIn-houseQueued
Status and results return to the candidate record they belong to.
Proof, not promises
We freeze our predictions. Then we grade them.
Every benchmark challenge seals its gold answer up front. The system commits a frozen prediction against a holdout it provably could not have seen — and the evaluation is published component by component, misses included.
- SEALED
The answer is sealed
The gold holdout is locked before any prediction exists.
- FROZEN
The prediction is frozen
The ranking is committed, immutable, against the sealed holdout.
- UNSEALED
The holdout is unsealed
Data the system never saw enters the record.
- GRADED
The evaluation is published
Candidate, context, mechanism, phenotype — each component scored, each miss labelled.
Every claim on this page describes shipped behavior — checkable in the product, never a roadmap.
Continuity
Runs end. The record should not.
Decisions, protocols, evidence, and reports stay connected to the work that produced them — so a result is still explainable months later, by someone who did not run it.
Getting access
Access is deliberate, by design.
CIRRA handles sensitive research data, so every lab is provisioned by a platform admin — never self-serve.
- 01
Request access
Tell us about your lab and research focus.
- 02
Admin review
A platform admin reviews every request before any workspace is created.
- 03
Lab activated
Approve the emailed link and your governed workspace is provisioned.
Already have an invitation? Use the link in your email, or sign in.
Keep the evidence, the decision, and the validation plan together.
Request access and a platform admin will review your lab and provision your workspace.