We accelerate drug discovery
Pharma R&D efficiency is decreasing steadily, while the cost of new drug discovery is increasing exponentially. Vast amounts of structured data are being generated by the use of high throughput technologies in drug discovery/development. However, the analysis and interpretation of this data for meaningful outcomes has been a challenge.
Peptris has developed a platform technology to enhance efficiencies across the drug discovery/development cascade using Artificial Intelligence/Machine Learning. Our neural network models aid us to rationalize and provide insights to experimental data in molecular biology and, further make predictions to generate hypothesis for newer experiments to be done
Our unique platform technology is not specific to any target, and aims to address a wide range of problems from predicting individual properties to analyze system interactions, including scenarios involving novel targets. The time, effort and money spent on new drug discovery can be optimized with the products that we are developing.
Predict binding affinity of small molecules targeting proteins for therapeutic purposes
Predict interaction scores of novel drug like peptide with target proteins
Re-purpose existing licensed drugs for new medical indications, since information about pharmacology, formulation and potential toxicity are available, drug development requires much lesser effort and can potentially reach clinical trials quickly
Our technology leverages cutting edge neural network architectures found in Natural Language Processing and Image processing research areas. We design and develop deep neural network models to learn from the vast amount of existing knowledge about proteins and small molecules. Our proprietary models are custom trained on a multitude of tasks specially designed to find generalized patterns and semantic representations of proteins and small molecules.
We have curated millions of data points including structure, annotations and properties data of Proteins, Molecules and Interactions from public data sources.
The learned representations of Proteins and Molecules from our pre-trained Multi-task networks enable us to predict binding affinities of protein-small molecule and protein-protein interactions. This fundamental tool makes it possible to accelerate drug discovery process,(discover,lead optimization,toxicity and specificty predictions) with unparalleled precision and accuracy.
We use very diverse set of data to train our models, and hence can be used to predict variety of properties of biological molecules and also for novel targets which donot have 3D structure. Such wide applicablility also enables prediction of selectivity and off-target toxicity
Our computational models work on cloud machines and deliver much faster than ultra high throughput screening. we can run our models to screen millions of compounds each day, and approach is fast enough to ensure screening for specificity as part of the pipeline.
Our deep neural network models provide far better accuracy than the current industry standards and we’re able to deliver accuracy comparable to wet lab experiments.
Our method is complementary to existing industry practices, and hence can be adopted very easily in the existing drug development process.
info@peptris.com +919945274057
Bangalore