Reproducible
Inputs, models, versions, outputs, and decisions remain attributable so an analysis can be reconstructed rather than merely repeated.
Computational Oncology
Computational infrastructure for personalized neoantigen vaccine design.
NeoCompiler makes personalized neoantigen selection reproducible,
testable, and scientifically accountable.
It preserves the scientific chain of custody from patient molecular data through prediction, scientific review, target selection, and downstream vaccine design.
The central idea
The prediction is only part of the decision.
NeoCompiler treats the full selection process as a scientific system. Prediction models can be evaluated independently, compared for agreement and disagreement, and tested under protocols fixed before biological outcomes are known.
01Patient Data
02Candidate Evidence
03Prediction
04Scientific Review
05Selection
06Vaccine Design
Every decision remains traceable.
Scientific architecture
Built for evidence that can be challenged.
Blindable
Selection rules can be fixed before biological outcomes are revealed, allowing performance to be tested without changing the rules after seeing the result.
Modular
Prediction models can be independently qualified, compared, activated, or substituted without redesigning the scientific workflow.
Traceable
Selected targets remain connected to the molecular evidence, computational evidence, and scientific decisions that produced them.
Validation
Scientific authority is earned in stages.
NeoCompiler separates software integrity, predictor qualification, controlled execution, and biological validation into distinct scientific gates.
Scientific collaboration
Built for blinded comparison, not retrospective storytelling.
NeoCompiler supports retrospective blinded validation and prospective research studies in personalized neoantigen vaccine development.