We provide several state-of-the-art in silico solutions for our partners working with proteins, antibodies and other biologics.
Our solutions are powered by the latest advances in AI/ML and enabled by our scalable cloud-based platform.
We keep abreast with the current state of the art in AI/ML and protein engineering and develop and integrate the latest and greatest technologies into our platform and process.
We incorporate the latest advances into our in-silico solution, ranging from graph neural networks, convolutional neural networks, diffusion models and other deep learning techniques, achieving state-of-the art results across a range of protein engineering problems.
Below we describe some of our partnering programs and protein engineering problems we solve using our technology. Using our technology, we can also provide custom solutions for your drug discovery challenges.
In the absence of structural data, we perform sequence-to-structure predictions for all proteins and biologics. As part of characterisation, we predict protein-protein interaction sites on the surface of proteins in aid of identifying the mechanism of action of a biologic. When working with antibodies we predict residues that comprise the paratope and use this information to make informed predictions about potential epitopes. If multiple antibodies are to be analysed, we perform epitope binning and determine epitope overlap.
PPI site prediction
When developing a novel biologic for a target protein, we perform structure-based virtual screening of our structural libraries in order to identify potential hits with preliminary complementarity in shape and chemistry. Alternatively, we design libraries of de novo protein scaffolds using the structure of the target protein. This process can involve a predefined region or epitope on the target protein or a predicted PPI site or epitope.
De novo binder design
De novo antibody design
Our proprietary binding affinity optimisation process utilises structural information of the drug-target complex, taking into account amino acid mutations themselves as well as their surrounding environment. If the goal of a project is to improve the species cross-reactivity of a biologic for studies in animal models, we provide simultaneous affinity maturation based on several homologous target proteins at once. We de-risk the affinity maturation process by filtering out mutations that would deleteriously affect the solubility or stability of the biologic.
Cross species reactivity
In instances where biologics have been obtained from non-human sources, we analyse the sequence and introduce mutations that resemble the closest human-germline reference sequence. Alternatively, we graft regions of the biologic, such as the CDR loops, onto a human framework. In these processes, we increase the “humanness” of the sequence with the goal of reducing immunogenicity. We validate that the changes introduced do not have a deleterious effect on the structural integrity of the biologic by predicting the structure of the newly developed sequence.
Humanised structure prediction
We consider key characteristics of candidate biologics in order to mitigate downstream developability issues. We assess the structure of the candidate in order to quantify features such as hydrophobic and charged patches; key factors relating to aggregation and colloidal stability. We further assess the sequence of the candidate for post-translational modification liabilities which can result in downstream issues during the drug development process.
And other PTMs
Eve is our cloud-based software platform for biologics design, discovery and optimization. It integrates our machine learning models, workflows and pipelines with visualisation and project management functionality. We use Eve internally across the various partnership programs we provide.
Eve was built to operate at speed and scale. Following best practices for software development, optimised compute infrastructure, data management and security, Eve enables the rapid development of biologics-based therapeutics.
We know that turnaround times for drug discovery campaigns are critical for success. Eve is powered by a distributed network of compute infrastructure, running on the latest CPUs, GPUs and machine learning frameworks. Our algorithms are optimised to be fast and efficient, delivering better results faster.
Whether you are working on one drug pipeline or a hundred, Eve was built to scale. By utilising the distributed and scalable nature of cloud computing, we can support hundreds of drug discovery pipelines simultaneously.
The security of biological data, structures and sequences is a top priority for us. Eve implements the latest security standards, encryption technology and access control mechanisms, ensuring the security of your data.
Copyright © 2021-2023 SilicoGenesis. All Rights Reserved.
Leuven, Belgium and Johannesburg, South Africa.