We propose a novel in silico drug discovery approach to identify kinase targets that impinge on nuclear receptor signaling with data generated using high-content analysis (HCA). Using imaging-derived descriptors, we provide prediction results of drug-kinase-target interactions based on single-task learning, multi-task learning, and collaborative filtering methods. These promising results suggest that imaging-based information can be used as an additional source of information for existing virtual screening methods, thereby making the drug discovery process more time and cost efficient.

Cancer is a deadly disease that causes an estimated 9.6 million deaths a year. Pharmaceutical drugs are important but developing new drugs is difficult and expensive. Here we generate a new peptide for PD-1, which is closely linked to a wide variety of cancers, using a new application called GANDALF to design new peptides. We present a peptide generated by our prototype to bind with PD1 and compare it to FDA approved drugs and results from a comparable method, Pepcomposer.

Standigm has applied deep generative models to design novel therapeutic compounds and launched Standigm BEST®, a proprietary molecular generative platform for lead discovery and optimization. On top of the main molecular generative algorithm, we developed an automated molecular design workflow to optimize and prioritize machine generated compounds for further synthesis and experimental validation. The most recent progress including real-life case studies will be shared.

Decisions taken in early drug discovery, from target selection to selecting the right chemical series greatly impact late stage attrition. We have developed data driven workflows that integrate heterogeneous data for target selection, lead identification and optimization in a holistic manner.This leads to better predictability and targeted experimentation.
 

This presentation will focus around the role of life sciences big data, technologies and emerging application of AI for the early drug discovery. The case studies discussing the impact of automation and miniaturization approaches coupled with machine learning on the speed and efficiency for the compound screening will be discussed. Additionally, a brief assessment of the therapeutics area-wise activity, current trends will be provided.

Computer models that are capable of predicting several thousands of biological activities for any chemical along with their ADMET properties have improved dramatically with the rapid growth of experimental data. The resulting network, illustrated by cancer drugs, has an extensive multi-target profile for each drug. These models use different mathematical methods, and help to predict new targets for known compounds, repurpose to new indications, search for compounds with specific multi-target profile, or identify potential liabilities.

Easy; Secure; Collaborative.  As the industry’s most trusted cloud-based drug discovery data management platform, CDD Vault has provided a secure, performant solution for early-stage research informatics for over 15 years.  Now, CDD has brought that same philosophy and expertise to the field of semantic drug content data and assay metadata with our new BioHarmony platform.  Speed up your drug discovery process with structured FAIR data - standardized, up to date, and ready to use

We have been building artificial intelligence models of metabolism and reactivity. Metabolism can both render toxic molecules safe and safe molecules toxic. The artificial intelligence models we use quantitatively summarize the knowledge from thousands of published studies. The hope is that we could more accurately modeling the properties of medicines, to determining whether metabolism renders drugs toxic or safe. This is just one of many places where artificial intelligence could give traction on the difficult questions facing the industry.

Better data is an obvious need for improved AI predictions, but have you considered the role of chemical descriptors?  See how better descriptors improve predictions across multiple algorithms. This study by Dr. Alpha Lee (University of Cambridge) was peer reviewed and accepted at the Journal of Chemical Information and Modeling.