Silica Corpora utilizes a proprietary AI platform to design and optimize therapeutic antibodies by analyzing amino acid sequences, enabling the rapid generation of high-quality candidates for priority diseases. This approach significantly reduces the time and resources needed for drug discovery, addressing the challenges of escalating costs and lengthy development timelines in the pharmaceutical industry.
Funding
Funding not disclosed
Founders
Product
Problem
The traditional process of therapeutic antibody discovery is often slow, expensive, and requires extensive experimental validation, hindering the rapid development of treatments for diseases with unmet needs. Existing methods may struggle to efficiently identify high-quality antibody candidates with desired properties from vast sequence spaces.
Solution
Silica Corpora offers an AI-driven platform for de novo design and optimization of therapeutic antibodies, accelerating the drug discovery process. The platform leverages proprietary protein Large Language Models (pLLMs) trained on amino acid sequences and project-specific data to generate novel antibody candidates. Its modular system allows for precise control over antibody properties, enabling the creation of candidates tailored to specific target product profiles. By integrating data from in vitro workflows, the platform achieves high accuracy with minimal input, reducing the need for extensive experimental screening.
Target Audience
The primary target audience includes pharmaceutical and biotechnology companies seeking to accelerate antibody discovery, improve candidate quality, and reduce development costs.
Features
- AI-powered Generator module that delivers antibody candidates as amino acid sequences of heavy and light chains, using only amino acid sequences as input.
- Discriminator module that predicts antibody properties such as developability, thermal stability, solubility, self-aggregation, and immunogenicity.
- Optimizer module that adapts antibody candidates to meet program-specific target product profiles (TPPs) by generating mutated amino acid sequences with predicted characteristics.
- Ep-Mapper module that accurately predicts epitopes on the antigen through both antibody-independent and antibody-dependent prediction for linear and conformational epitopes, without relying on 3D structures.
- Holistic design approach that considers all relevant parameters simultaneously to create successful antibody candidates.
- Models trained on millions of unique protein-protein interactions and fine-tuned on thousands of antibody-antigen interactions.