Lu Rahman spoke to Ashley Zehnder, CEO of Fauna Bio, a biotech harnessing the power of comparative and functional animal genomics to find better therapies for human disease, including neurologic, metabolic, fibrotic and cardiovascular disorders.
“Comparative genomics is the process of using data from mammals and other species to show us where the most important genetic sequences lie and can help us better understand the genetic basis of human disease,” says Ashley Zehnder CEO and co-founder of Fauna Bio. “When exploring comparative genomes, it’s important to understand that we share about 90% of our genes with our mammalian cousins, many of which have evolved in unique ways to survive in challenging environments. There is widespread agreement among researchers that the comparison of genomes, from the more than 6,000 mammals that inhabit our planet, will help geneticists identify key highly conserved sequences, which are much more likely to contribute to disease.”
At the core of Fauna Bio’s process is a precisely timed biobank of samples from hibernating animals, which provides genetic, epigenetic and proteomic data. “Hibernating animals survive under extreme conditions, possessing mechanisms of self-healing and tissue repair to endure long periods of time in freezing temperatures without food,” explains Zehnder. “Furthermore, hibernators become obese while readying themselves for being immobile and eating nothing for months, and they survive without developing diabetes, muscle atrophy or neurological damage, despite experiencing physiologic changes that closely resemble human heart attacks and stroke.”
Fauna Bio’s platform navigates to the genes responsible for these traits between mammals, including humans, to find the specific genes that can resist disease. Zehnder says: “These genetic pathways help us target the conserved genes that already exist in humans to generate novel targets and develop therapeutics to help overcome diseases that devastate human health. To date, the platform has identified two compounds and four novel genetic targets currently undergoing evaluation for heart disease and related disorders as well as seven novel genes and six compounds that can reverse fibrosis in human cell models. The company’s goal is to initially develop therapeutics that can be partnered with pharmaceutical companies once they complete preclinical or early clinical stage development, meanwhile building the capacity to have internal assets progress to FDA approval.
Using machine learning
Fauna Bio describes itself as addressing a $60-plus billion potential market with first targets in heart disease, stroke, Alzheimer’s and diabetes. “We are using machine learning (ML)techniques to analyse proprietary data from the world’s most precise hibernator biobank, hundreds of additional datasets from more than 30 species and specific expertise in genomics, transcriptomics and physiology to identify disease resistance genes and develop drugs to treat human diseases,” says Zehnder. The company uses extreme adaptations in mammals to highlight key genetic nodes that can modify disease-causing pathways in people.
She explains that the difficulty in identifying highly effective drug targets stems from the difficulty of separating genes that are causing disease from those that are ‘just along for the ride’ or are secondary changes. “This is really hard to do using human data alone. Fauna’s platform can highlight genetic networks that are different between the protected state of certain species – eg hibernators– and the disease state of humans.”
“For example,” she adds, “we can examine what genes are changing in the hearts of hibernators at timepoints where they are protected from physiological conditions that resemble human heart attacks. We can then layer onto those data genetic networks from human patients with failing hearts and pinpoint exactly which genes to target. Using a trans-species approach allows us to essentially ‘geolocate’ the key human disease genes more rapidly and with higher precision. To help us, we have integrated data from OpenTargets, the UK Biobank, Reactome and several hundred purposefully curated functional genomics studies from humans and other, more traditional, model organisms.”
Fauna Bio uses ML to determine which of our metrics are the most important; it turns out that a mix of data is the key. “What we aim to see are the same genes coming up over again in diverse datasets that relate to the same disease. For example, in our current screening programmes for cardiac disease, fibrosis and obesity, we have a 20% positive hit rate on drug targets in vitroand 100% hit rate moving those targets in vivo that have positive in vitrovalidation. Our targets were identifiedin vivo(in hibernators), so they are more likely to be functionalin vivo,” says Zehnder.
The Fauna platform is an end-to-end disease to gene to compound discovery pipeline including data from 420 species, millions of gene-disease linkages, and 34,000 small molecule compounds. “Our LEO module enables us to directly predict new uses for small molecules and the data sources for that have increased from around 40,000 perturbations and 1.5 million gene expression profiles to 80,000 perturbations and three million gene expression profiles,” says Zehnder.
“Part of what helps us is the reduction in sequencing costs over the past 20 years. The cost of assembling the first human genome was close to $3 billion. Now a new genome assembly for a species can be done for close to $10,000. Similarly, costs for RNA-sequencing, proteomics, epigenetics have plummeted, making multi-omics studies in diverse species possible at a scale never before possible,” she adds. Due to this dramatic drop in cost, the amount of available sequence data Fauna Bio has on vertebrate animals dwarfs the number of available human genomes. “As of 2019, there was functional genomics data available on over 30,000,000 animals and this number is growing exponentially, with current sequence archives containing data from > 600 different species. All together, less than 1% of the available vertebrate sequencing data available today is from humans,” Zehnder explains.
Addressing the challenges
“We’ve had the full sequence of the human genome for nearly 20 years and in that time, the major disease killers of humans have remained unchanged,” says Zehnder. While overall drug approvals have increased in recent years, more than three quarters (78%) of FDA-approved drugs in 2020 and 2021 act through known targets or pathways. Zehnder says: “At the core of the problem is that, overall, humans are just too similar to be able to use human data alone to tease out the most important functional networks. Our CTO, Linda Goodman, encountered this problem while working on her PhD looking at variations in the human genome linked to heart disease. While the variants she discovered were statistically significant, their overall effect on actual disease in humans was very small. She recognised that a better way to pinpoint disease causing variation was to examine which genes are conserved evolutionarily over 100s of millions of years.”Fauna recognises that these same highly conserved genes are functional in a range of species, many of which use them to protect themselves from disease.
“By leveraging extreme variation across mammals, Fauna Bio can accentuate the role of conserved genes, permitting the identification of new drug targets and compounds with a much smaller number of samples, and thus, with much lower cost,” says Zehnder. “While many GWAS studies can use <10,000 participants, some larger studies have used more than a million participants for complex diseases such as Alzheimer’s disease, psychiatric disorders and heart disease.” The Fauna platform has identified functional targets using heart data from only 60 individuals in its biobank at timepoints where it is known that the animals are able to protect themselves from damage. “Reducing the expression of these genetic targets doubles the survival of human heart cells treated with hypoxia (lack of oxygen) in lab experiments and almost completely prevents heart damage from surgical studies in rodents that mimic human heart attacks,” she adds.While there is a large number of drug discovery companies using ML, AI and genomics, most of these companies, says Zehnder, are using similar datasets (human genomics data, data from human or animal cell models or animal disease models). “ML platforms are only as good as their input data. Fauna Bio brings novel datasets to the era of genomics drug discovery and combines that with computational techniques to bring new therapies to large areas of unmet need in human health,” she adds.