SBIR Phase I: Artificial Intelligence Platform for Peptide Drug Discovery

这项小企业创新研究（SBIR）第一阶段项目的更广泛/商业影响是改进了肽疗法的药物发现过程。多肽药物的主要治疗领域是糖尿病、肥胖症和癌症。抗体通常是一种治疗选择，但价格昂贵，给患者及其家人带来了巨大的压力。合成多肽药物在具有相似特异性和低毒性的同时，生产成本较低。该技术可提高临床前先导药物的开发效率；节省20%的时间将在单个发现项目中节省420万美元。这项小型企业创新研究（SBIR）一期项目旨在展示和实验验证一种新型神经网络框架准确预测具有抗炎活性肽的能力。提出的创新解决了当前人工智能方法的基本限制，如应用于肽数据集，通过使用新的数据编码和创新的模型架构。该项目将涉及进一步开发架构，并结合有效的数据增强和迁移学习策略，以有效地利用小数据集。此外，模型预测的肽变体将通过生成湿实验室数据和评估实际表现进行广泛验证。将开发可视化有前途的肽模式的工具，以实现广泛采用人工智能方法所需的模型。该提案的成功完成将导致新的抗炎先导肽和独特的深度学习框架，以加速和改善肽药物的发现。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The broader/commercial impact of this Small Business Innovation Research (SBIR) Phase I project is an improved drug discovery process for peptide therapeutics. The key therapeutic areas for peptide drugs are diabetes, obesity and cancer. Antibodies are often a treatment of choice but can be expensive, putting significant strain on patients and their families. Synthetic peptide drugs can be cheaper to produce while having similar specificity and low toxicity. This proposal’s technology could improve the efficiency of pre-clinical lead development; a time savings of 20% would result in $4.2 M in savings in a single discovery program. This Small Business Innovation Research (SBIR) Phase I project aims to demonstrate and experimentally validate the ability of a novel neural network framework to accurately predict peptides with anti-inflammatory activity. The proposed innovation addresses fundamental limitations of current artificial intelligence approaches, as applied to peptide datasets, via use of novel data encoding and innovative model architectures. This project will involve further development of the architectures and incorporation of effective data augmentation and transfer learning strategies to effectively leverage small datasets. Moreover, the model-predicted peptide variants will be extensively validated by generating wet-lab data and evaluating real-life performance. Tools to visualize the promising peptide patterns will be developed to enable models necessary for widespread adoption of artificial intelligence approaches. The successful completion of this proposal will result in novel anti-inflammatory lead peptides and a unique deep learning framework to accelerate and improve peptide drug discovery.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.