DISQOVER for chemistry and early drug discovery

Challenges of early drug discovery

In early drug discovery, researchers need to access all available and relevant information about compounds. In this way, they can identify if similar compounds have already been tested and whether toxicities, adverse effects, etc. have been reported. However, accessing and analyzing all the required information across databases and data silos is difficult, time-intensive, and costly. Even more, there is a significant risk of overlooking important pieces of information, which could lead to decisions based on flawed or incomplete knowledge.

This is where a data-driven approach comes in: instead of manually checking if something has been reported before, the data scientist can use machine learning to automatically detect whether a specific adverse effect has been reported previously. This is possible because databases and other sources of information contain certain patterns that can be used by the machine learning algorithm as features.

A scientist could then train a machine learning algorithm on these patterns and let it identify the relevant features automatically. The trained algorithms can be applied to new datasets, allowing scientists to detect relevant features more quickly than ever before. This reduces the time scientists need to spend on manual analyses while increasing their productivity and making them more efficient in their work. The efficiency gains here are huge!