DrugnomeAI

About

DrugnomeAI is a machine learning tool that has been developed in aim of generating a target-druggability scoring for the entire human exome. Understanding druggability is important, as this can suggest higher efficiency in identification and validation of potential drug-targets – particularly in earlier stages of the pipeline. Furthermore, this can help accelerate the prioritisation of novel drug targets.

Challenge

With only 10% of the human genome representing druggable targets, and only half of those being relevant to disease, it is important to be able to predict how druggable a novel target is in early stages of drug discovery pipeline.



Opportunity

Use an existing ML gene prioritization framework (Mantis-ML) to address a new type of problem – gene druggability.



Aim

Create a generic druggability profile across the entire exome.

Utility

  • Identify novel drug target opportunities
  • Increase efficiency of target identification and validation
  • Drug re-purposing: enable tackling still untreated diseases
  • Expand our understanding of druggability by illuminating gene properties that predispose their binding affinity

Getting started

Query genes in the search field at the top to explore their druggability rankings according to DrugnomeAI. Different models of druggability are provided, some are generic and some are modality specific.




Genes are ranked according to their raw druggability score - which varies between 0 and 1. The raw score can be thought of as the 'probability a gene is druggable' under a given model. Over 19,000 genes are then ranked according to these scores, with ranks normalised to be out of 100 (this is referred to as the 'percentile of score'). All scores are available to download from tables - along with their rankings.