Deeper than validation: How Verge Genomics establishes causality for its drug targets

Much of the attention Verge Genomics has received has focused on our artificial intelligence (AI) and machine learning (ML) techniques to identify drug targets for the most challenging diseases of our generation. While these techniques differentiate Verge in the space of neurodegenerative disorders, the investigatory work we do in therapeutic discovery also sets us apart.

Verge’s all-in-human approach starts with large cohorts of high-quality brain tissue samples from patients and healthy subjects. Using the ML approaches of our CONVERGE™ platform, we identify dysregulated gene networks that point to the mechanisms of neurodegeneration. Our platform then synthesizes a wide range of systems biology data to identify so-called master regulator genes that are predicted to therapeutically restore the activity of disease networks to healthy-type levels. We show that targeting these regulator genes with our novel drugs restores the neuronal health in human and animal models of neurodegeneration. 

A very important step between computational target discovery and developing a new therapeutic program is a process called “target validation”. In target validation, we establish the causal role of the gene in human disease and demonstrate that drugging the target could benefit patients. Success in target validation will trigger the next step: the start of drug discovery.

In this article, I’ll describe how CONVERGE™ goes beyond conventional validation and accelerates our progress towards the clinic. 

The case for causality

How does Verge Genomics use the CONVERGE™ platform to identify new drug targets? We draw on integrated technological innovations across computer sciences, biology, chemistry, and translational research to arrive at the answer. 

First, we developed proprietary in-house computational algorithms to identify gene networks dysregulated in the brains of patients. CONVERGE™  identifies gene targets by analyzing a wide range of ‘omics’ data, including the transcriptome, the protein-protein interactome, knowledge of gene regulation, and statistical genetics approaches of expression quantitative trait loci (eQTL) and Mendelian randomization. We find that many networks are linked to genetic drivers, for example TPD-43 or FIG4 in amyotrophic lateral sclerosis (ALS), providing a strong indicator of the causal role of the target in disease.

Second, we computationally link our expression data for cohorts of tissue samples to a rich catalog of clinical data we have collected on those cohorts: patient genotype, demographics, age of onset, pathology, disease progression, and more. This analysis identifies gene networks – and their master regulator targets – that are implicated in disease onset, progression, and severity.

Third, we probe the human tissue-derived target’s role by employing cellular models that replicate a patient's neurodegenerative features. We generate these human models from patients’ skin cells, which are then coaxed into developing into neurons or other brain cell-types, a method for which the Nobel Prize was awarded in 2012. In doing so, we have created an “all-in-human” approach that we believe increases the probability of translational success by going from human to human rather than animal to human. 

What emerges from the CONVERGE™ platform is a deep understanding of the target’s causal role in disease biology, providing the catalyst for developing new therapies: our PIKfyve inhibitor is poised to enter clinical testing for ALS at the end of this year, only four years after the initial computational discovery.

Targets off the beaten path 

Some targets might be linked to a specific pathway, such as lysosomal trafficking in ALS in the case of PIKfyve. Other targets may have multiple roles, or their function is less obvious, or they have received little or no attention in the disease. In these cases, CONVERGE™ determines which phenotype is the most relevant to rescue in the disease. 

Over the last six years, we have built a comprehensive and well-curated biorepository of human brain cell types, derived from patient skin cells. Through target validation and drug discovery work across sporadic and genetic forms of disease – such as C9orf72, TDP-43 and SOD1 mutations – we built up a clear picture of the patient cohorts that were most likely to respond to treatment, de-risking the development of our PIKfyve program.

We have also built a comprehensive suite of sophisticated tools to probe target activity by leveraging new advances in biological engineering, including genetic reagents such as CRISPR or small molecule compounds. Our chemists design and synthesize novel, biologically active chemicals, providing an entrée into drug discovery. We perform broad sweeps of disease phenotyping of our patient-derived cell cultures after interfering with the activity of the target, integrating transcriptomic and proteomics network analysis with single cell tools, electrophysiology and deep imaging of dozens of cellular endpoints. It is this unique panoramic view of target biology that drives our ability to demonstrate prevention or reversal of pathology.

Opportunities for networking

What sets Verge Genomics further apart in the quest for finding cures for patients is how we use our computational platform in all aspects of the therapeutic pipeline, from the initial discovery of the target all the way into the clinic. 

Many drugs fail in the clinic because preclinical models lack predictive power. We deploy in-house algorithms and ascertain the best experimental models, looking for evidence that the patient’s dysregulated gene network is also present in the cells that we use. It is exciting to see that we can restore these diseased networks to the healthy state after application of our drugs, results which spearheaded Verge’s lead programs in ALS and Parkinson’s disease.

CONVERGE™ allows us to discover targets that others don’t find or don’t recognize as important in the disease. Because we invest the upfront effort in gaining deep insights into target biology, and in finding ways to reverse it, we prioritize disease targets that make sense to us. This derisks our development of better, more effective drugs that patients with neurodegenerative diseases need.