Frances Pearl, Senior Lecturer in Bioinformatics, University of Essex outlined some of the techniques she uses to identify drug targets at the Sussex Drug Discovery Group. Upon joining Sussex and helping establish the Sussex Drug Discovery Centre, Pearl’s team began identifying drug targets for a Welcome Trust grant, leveraging the Genome Damage and Stability Centre’s expertise. 

Pearl’s presentation focused on using a synthetic lethality approach that targets a gene that is essential only in cancer cells without affecting normal cells. For example, patients with BRCA1/BRCA2 deficient breast, ovarian, and pancreatic cancers are usually given Olaparib which doesn’t target the BRCA gene but targets its synthetically lethal partner, a PARP inhibitor.  

Several methods have been developed to look at synthetic utility. Slorth was the original one used by Pearl, it was produced before there was any extensive cancer data. So, they looked at five different model organisms: two types of yeast, flies, worms, and humans. In humans, only 150 genes were reported, which is a small data set. AI was used to predict synthetically lethal pairs. However, this method performed poorly at predicting synthetic pairs of proteins in humans but was effective in other organisms. 

Another method relies on an in-house programme that determines whether a missense mutation is linked to function gain or function loss. To figure out whether something has a loss of function the team relied on DNA methylation data rather than gene expression.  This method also adopted a multiomics centric approach by considering copy number and integrated mutations to identify synthetic lethal relationships. 

DepMap is a resource that searches for essential genes, it consists of CRISPR knockout data for over 1,000 cell lines. DepMine analyses the DepMap data to identify synthetic lethal relationships. 

To demonstrate this strategy in practice, Pearl showed the essentiality of beta-catenin in different cell lines. She said: “In normal cells, when you've got a wild type APC gene, beta-catenin here is not essential and you can knock it out and that doesn't matter. As soon as you knock out APC, which is this match set here, beta-catenin suddenly becomes essential. So that means in APC, in cells without APC, beta-catenin is a great drug target.” 

The latest method, DependANT, uses perturbed protein-protein interaction (PPI) networks tailored to individual cancer cell lines. By taking into account gene loss and expression changes, the model predicts essential genes more accurately. Pearl added that it can be trained on one tissue type and predict essential genes in different tissue types. To round up, Pearl mentioned her role in the CRUK initiative. She aims to make cancer data more accessible for researchers and support data sharing and reuse.