New York – Biotech startup Carta Biosciences is betting that its human gene interaction mapping technology can help drug companies reduce inefficiencies in their drug discovery efforts and more quickly discover novel drug targets.
Carta uses a systems biology approach and artificial intelligence algorithms to model and analyze causal pathways in disease with an eye towards identifying therapies that can reverse these pathways. The company believes that its approach will ultimately cut in half the current time and cost of drug development using traditional approaches.
Though Carta claims that its algorithms can work for any disease type which has an underlying gene expression signature, it focuses on conditions that affect the brain. Its focus areas are neurological cancers and specifically glioblastomas, aging-related conditions that affect hair and skin, and conditions that affect cognition, such as Alzheimer's, dementia, and epilepsy.
"A big issue for pharma is the requirement to show improved efficacy over existing treatment. It's not good enough just to come up with another drug for another disease. You've got to show that drug is better than the existing therapies," Michael Johnson, a professor of neurology and genomic medicine at Imperial College London and Carta's CSO and CMO, said in an interview. "The way that we've approached this is to model the disease state computationally at the level of gene expression data, and then you use systems biology approaches to identify new drug targets, novel drug targets that can be used to shift that disease gene expression signature away from the disease state and back towards the healthy state"
Carta believes that its computational approach improves on traditional approaches such as high-throughput screening. Its Founder and CEO Chee Yang Chen explained in an interview that the company's discovery engine requires high-quality single-cell gene expression, epigenetic, and proteomic data as input. It uses this data to map gene interaction networks within cells. It then compares maps of diseased and non-diseased cells to identify causal pathways for disease, Chen explained.
Details of Carta's computational engine are provided in a paper that was published in Nature Communications last year. That paper describes efforts to apply the company's method to identify new treatments for epilepsy as part of a partnership with UCB Pharma.
As explained in that paper, Carta's so-called Causal Reasoning Analytical Framework for Target Discovery, or CRAFT, "combines gene regulatory information with a causal reasoning framework to computationally predict cell surface receptors with a direction-specified influence … over disease-related gene expression profiles." In the study, the researchers used the method to predict a tyrosine kinase receptor, Csf1R, as a potential therapeutic target for treatment from disease-related gene expression data. The partners also identified several other candidate regulators of epileptic networks that they believe could also serve as drug targets for epilepsy.
Though some pharma companies are considering newer methods for drug discovery, many still work with outdated models for drug discovery, Johnson said. Currently, "there are still a lot of therapeutic targets which are derived from reductive scientific approaches and animal models or small molecule screening."
Chen pointed to the billion-dollar price tag that is currently required to develop new drugs and the fact that many small molecules don't make it market. The challenge is that "we do not understand what is going on; we are throwing everything at the wall [to] see what sticks," he pointed out. Not only is it a waste of resources, "what we are getting out of it are drugs that do not work on a causal basis and therefore provide hardly any improvement over current gold-standard treatments."
With its computational engine and in-house skills and expertise, Carta can help drug companies derisk their existing portfolios because "we can understand biology from its most fundamental point," he said. These companies "have a lot of things on their shelves, but they don't know what works and how to actually deploy their resources. We can help them with that."
Carta, though, has its sights set on discovering and developing its own drug targets as well, Chen said. "We really want to identify our own drug targets and that may be in collaboration with pharma," Johnson added. "The point at which we interact with pharma can be very early stage or it can be potentially later stage."
In terms of its business model, Carta forms bespoke partnerships with potential partners that are driven by their drug development needs. Chen said that his company is currently in discussions with multiple companies to iron out the details of their working relationships, though Carta is not identifying any of the companies by name at this time.
The exact nature of the partnerships that Carta forms depends in part on these companies' policies that govern their ability to access and share internally developed high-quality regulatory datasets, Chen said. Some companies are willing to share data and to co-develop any potential candidates with Carta. Meanwhile, Carta also plans to build its own internal datasets which it can use in its partnerships. It is currently creating at least one such dataset for glioblastoma, according to Chen.
Carta's pricing is flexible and depends on the nature of the agreement between the two companies, Chen said. For example, "the data that we are doing in GBM right now, that would be our own data, and obviously the price point would be much higher than if we went to a pharma company and they provided a lot of the data," he explained.
Carta points to companies such as BenevolentAI and Verge Genomics as competitors in its space. BenevolentAI, for example, also offers an artificial intelligence-based platform used for drug target identification among other tasks. In September, that firm announced that it had raised $90 million from Temasek, an investment company headquartered in Singapore, that it is using to further develop its platform for drug discovery and development.
"The selling point of our company is being able to go from first principles using biological data that is empirical and context specific," Chen said. "So, [for example,] we are looking at single-cell mRNA sequencing as opposed to bulk mRNA sequencing." In other words, "[we] reflect biology a lot more than all the other models."
Other factors that set Carta apart include its focus on neurological disorders and conditions, Chen noted. Also, Carta's engine does not rely on natural language processing methods to find relationships between drugs and diseases as some other approaches do, he said. Since NLP methods rely on parsing the literature, they may be relying on research that is not replicable and contains biases that affect their outcomes.
In addition to UCB Pharma, Carta has collaborations with Cancer Research UK, Brain Tumor Research, and Parkinson's UK. Carta plans to announce a round of funding in 2020.