Redefining Drug Development Timelines: The Silica Corpora Approach

Silica Corpora is at the forefront of transforming drug discovery through cutting-edge technology. In this interview, Jaime Rosselló, the CEO, explains how their language models specialized in protein-protein interactions are fundamentally changing the processes of drug research. Rosselló shares insights into the founding of Silica Corpora, their innovative solutions, and the company's future vision. Discover how Silica Corpora is paving the way for faster and more cost-effective drug development through tailored AI solutions.

How would you explain your solution in three sentences to a professional colleague in a chemical company?

"Large Language Models (pLLMs) trained on millions of unique protein-protein interactions and fine-tuned on thousands of antibody-antigen interactions, making a set of holistic protein information. The ability to infer structural information from sequence data is a significant advancement, as it sidesteps the need for direct structural determination for each protein of interest."

What problem motivated you to start the company?

"Silica Corpora emerged from a shared vision to transform drug discovery through cutting-edge technologies.

Tim and Jaime, former classmates turned co-founders, embarked on a mission to address industry challenges head-on. Their quest gained momentum with the addition of Anna, our exceptional Chief Scientific Officer.

With extensive expertise in the medical field, Anna boosted Silica Corpora's capacity to develop tailored AI solutions. Driven by a pressing need, the trio united to confront the escalating costs and time associated with bringing new prescription drugs to market. Silica Corpora was born out of a desire to rewrite this narrative."

The Silica Corpora Team

How do you convince a chemical company to set up a pilot project with you?

"That's why we are looking for long-lasting partnerships, to develop tailored solutions based on your needs.This ensure the possibility of building internal assets without having to deal with a simple contractual interaction, but a more trust-based cooperation that ensures IP protection and developability of trust.

To demonstrate the potential of our platform we are open to run a blind test of our epitope-specific binding prediction module.

The input for the test are Amino acid sequence of a target (including epitope) and Amino acid sequences of antibodies characterized as binders or non-binders for the target epitope (REMINDER: The labels of which antibodies are binders (and which are not) must be hidden from us to ensure the test is indeed blind!)

The Output will be the Probability of each antibody to be a binder for the specified epitope of the target. We provide the results in less than 30 days."

Who are your current customer (groups) and who do you want your potential customers to be?

"Our customers are any biotech company operating with Antibodies (or having discovery programs for antibody therapeutics) without strong AI capabilities. Ideally, those companies have discovery programs that are in desperate need of lead candidates for preclinical trials. Our first companies are active in Germany, but we are capable of working worldwide, focusing on US, Europe and Asia."

Where do you see yourself in 3 years and how can 5-HT support you?

"In 3 years our objective is to close our first target deal for antibody design with some of our long-lasting partners, including an improving of our De Novo design and the creation of our Foundation models for antibody design. 5-HT is a great partner to spread the word about our current achievments, and to help us bringing interested companies to develop our first proof of concepts that will lead to long-lasting collaborations."