Six Engines
Six organ systems of the digital lifeform · Autonomous & Controllable
If the digital lifeform is compared to an organism, these six engines are its organ systems-each with its own role, yet connected through the same pipeline bloodstream. The underlying technology benefits from the global open-source community. What DiVo Gen²AI does: deploy them as an autonomous and controllable service system, connecting them into a complete pipeline from DNA to digital life.
Engine Overview
Whole-genome variant annotation and VUS three-model cross-interpretation
Autonomy anchor: 636 pangenome
Protein 3D structure prediction (three-model cross-validation)
Autonomy anchor: 400+ complexes, pLDDT 95.4
3D multi-cell population simulation
Autonomy anchor: Self-compiled + personalized
Single-cell signaling pathway solver
Autonomy anchor: Self-deployed + pathway library
Immunogenicity assessment (MHC-I/II multi-allele)
Autonomy anchor: 65+25 allele, IC50 10.1 nM
mRNA sequence design and codon optimization
Autonomy anchor: Spearman 0.92, CAI 0.7->0.95
Three Layers of Autonomy
Not a slogan, but three concrete layers
Compute Autonomy
All six engines run on DiVo Gen²AI's self-deployed servers. Structure prediction, genome interpretation, immunogenicity assessment, sequence design-every compute node is local inference, not dependent on external API availability or pricing fluctuations.
Data Sovereignty
Chinese genomic data is protected by the Human Genetic Resources Management Regulations. Full-chain autonomous deployment means data never leaves the controlled boundary from input to output-not sent abroad, not read by third parties, not used to train public models.
Engineering Iteration Autonomy
560K lines of self-developed code-when science has new discoveries, models have new versions, or data has new dimensions, DiVo Gen²AI can autonomously update pipelines. No need to wait for upstream vendor scheduling.
Verifiable Engineering Foundation
Not a vision on a PPT-already running, validated, and deliverable engineering capabilities
| Capability Dimension | Validation Metric |
|---|---|
| Protein structure prediction precision | pLDDT 95.4 / ipTM 0.977 |
| Batch protein complex prediction | 400+ |
| MHC-I binding prediction | 65, IC50 10.1 nM |
| MHC-II binding prediction | 25 (13 DR + 5 DP + 7 DQ) |
| mRNA sequence design | Spearman 0.92 |
| Chinese population genome optimization | 636 |
| Neoantigen pipeline | HLA -> variant -> MHC -> pMHC -> immunogenicity -> mRNA |
| Self-developed code scale | 560K lines |
| Model coverage | 41+ |
| End-to-end pipelines | 8 (S1-S6 + R1-R4) |
| Model distillation capability | 744B -> 60B |
| Data compliance | Full local processing |