1
Observed
ZNF133 was silenced in HepG2 cells
A measured CRISPRi gene-expression response is available in the released same-context dataset.
Open the observed response →Evidence Graph
Follow one claim from observation to the missing experiment.
The graph shows why NMD-VCell can recommend a study without claiming that a gene will improve DMD.
Worked example
ZNF133 evidence path
Each step uses one of three ordinary-language states: observed, limited context or not yet tested.
2
Limited context
ZNF133 was checked in a human-myoblast screen
It was represented in the screen but did not meet that screen’s hit threshold. This is context evidence, not DMD replication.
Open the gene evidence →3
Evidence disagrees
DMD datasets do not fully agree on direction
The disagreement is retained as uncertainty instead of being collapsed into a therapeutic claim.
Inspect the DMD evidence →4
Not yet tested
No independent ZNF133 perturbation exists in DMD muscle
This is the decision-blocking gap. The absence of an experiment is not a negative result.
See why prediction is locked →5
Next experiment
Test matched DMD and control muscle cells
Use two independent reagents, biological repeats and one prespecified functional outcome.
Design this experiment →What this means: the platform does not turn several indirect clues into a confidence score. It preserves the chain, exposes the missing link and sends the user to the experiment that would resolve it.