01
AI Neoantigen Prediction
AI 新生抗原预测
Advanced deep learning architecture combining CNN, LSTM, and large language models with molecular simulation for superior prediction accuracy
结合 CNN、LSTM、大语言模型与分子模拟的先进深度学习架构，实现优异的预测精度
0
Cross-Species Prediction Accuracy
跨物种预测准确率
0
MHC-II Epitope PPV
MHC-II 表位预测性能
Algorithm Architecture
算法架构
CNN + LSTM + Large Model + AlphaFold & Molecular Simulation
CNN + LSTM + 大模型 + AlphaFold & 分子模拟
Structural Biology
结构生物学
MHC pocket constraints & TCR binding information
MHC 口袋约束及 TCR 结合信息
Multi-Species
多物种预测
Pigs, chickens, ducks with >80% accuracy
猪、鸡、鸭等多个物种，准确率 >80%
MHC Coverage
MHC 覆盖
Both MHC I & II with leading domestic performance
MHC I & II 均可准确预测，国内领先
Validation
验证
Multiple animal studies & ~100 clinical samples
多个动物实验及近百临床样本验证
Publications
已发表
Front Oncol., Science Advances (IF>17)
| Technical Parameter | 技术参数 | TheraRNA | 澄实生物 | BioNTech |
|---|---|---|---|---|
| Sequencing Technology | 测序技术 | 3rd Gen Long-Read | 全长三代测序 | 2nd Gen Short-Read |
| Capture Method | 捕获 | Translatome | 翻译组 | WES + RNAseq |
| Antigen Prediction | 抗原预测 | MHC I + II | MHC I + II | MHC I |
| Patient Response | 患者响应率 | 80% (4/5) | 80% (4/5) | 50% (8/16) |
| Positive Epitope Rate | 表位阳性率 | 19.6% (11/56) | 19.6% (11/56) | 10.4% (25/240) |
Data Sources
数据来源
Public IEDB data + ~1,000 internal tumor clinical datasets + Collaborative 1M+ immunopeptidome data
IEDB 公开数据 + 自有近千肿瘤临床数据 + 合作百万条免疫多肽组学数据
Computing Cores
计算核心
~400
CPU
Memory
内存
~32T
Storage
存储
~1000T
GPU Arsenal
GPU 阵列
H100
+ GTX 3090 & 4090
