CHICAGO – In responding to the COVID-19 pandemic, genomic interpretation software developer Breakthrough Genomics last month added a research-use-only coronavirus susceptibility assessment to its menu of services.
Irvine, California-based Breakthrough announced that it is leaning on its Enliter machine-learning and analysis platform to conduct SARS-CoV-2 susceptibility assessments by looking at genes that might influence susceptibility and resistance to the effects of COVID-19. The company said that Enliter is designed to "mimic the workflow" of US board-certified medical geneticists.
Enliter scours medical literature related not only to SARS-CoV-2, but other coronaviruses as well. Founder and CEO Laura Li noted that the novel coronavirus shares many DNA and RNA similarities with SARS-CoV-1 in particular.
"We scan all the literature and try to understand what are the factors involved in this risk as well as susceptibility," Li said. With the COVID-19 add-on, Enliter will run in silico studies to create genetic panels to predict risk based on available data.
Li, who founded the company in 2016 to provide genomic data analysis and interpretation services, said that Enliter scans new papers in real time, adding knowledge to its database to refine its predictions.
The company said that it has been able to identify at least nine genes and several related variants that it considers to be the "most influential" genetic factors in human response to coronaviruses in general and COVID-19 specifically. These genes include regulators of viral entry, replication, and destruction as well as those related to upper respiratory infections and both innate and adaptive immune response, according to Breakthrough Genomics.
"All of these steps [may help] determine how people are susceptible to the viral infection and whether the symptoms will be light or severe, based on all our understanding previously and also the new publications," Li said.
Initially, Breakthrough Genomics is offering the risk assessment as an add-on, research-focused module for current customers.
The announcement last month was meant in part to find partners to validate the technology and the in silico gene panel, COO Maribeth Raines said. She described the in silico test as "a theoretical panel."
Ideally, these partners would add to the knowledgebase that the machine learning relies on, Raines said. They also would collect data on whether there might be other genes indicating susceptibility risk that might be validated in a more formal prospective clinical trial.
"We're in many discussions with many different types of labs and businesses to look at that susceptibility," Raines said, but declined to elaborate.
Breakthrough Genomics is offering the assessment for research purposes only for current users of its whole-genome and whole-exome sequencing services and is in negotiations with others in the US and abroad, company officials said.
The firm is not ready to announce new partners or details about a planned clinical study, including a timeline for its COVID-19 next-generation sequencing panel.
One challenge Breakthrough Genomics faces is that much of its planned future work would be based on whole-genome sequences, though currently available data tends to come from whole-exome or microarray panels with targeted SNPs.
"That's the real challenge right now," Raines said. She said that the company is "in discussions for doing a very large study."
In the short term, Breakthrough is looking to partner with companies and healthcare organizations that already have WGS data, going back to individuals in those networks to check on whether they have the SARS-CoV-2 virus or COVID-19 symptoms.
"We already have the DNA," Raines said. "The issue is finding out what the COVID-19 status of those individuals are."
Breakthrough also is exploring how to work with patients who already have sequencing data from consumer genetics test like those offered by 23andMe and Ancestry.com. However, the company still has to tailor its technology to handle those reports and find someone to pay for the analysis.
Raines said that the company is capable of working with DTC test providers in addition to research organizations.
The DTC approach would require a more simplified panel, according to Raines. Li said that Breakthrough has started looking at 23andMe data for about one-third of the variants associated with COVID-19, but has not launched any such service.
"You don't get nearly the power that you get looking at that data versu looking at the whole genome, and there are a couple variants that are in introns, so we really prefer the whole genome over the whole exome," Raines said.
Breakthrough Genomics has what it calls Tier 1 and Tier 2 targets. The first tier, announced last month, has to do with viral processing and innate immunity, according to Raines. However, there is a second tier of about 15 or 16 human genes related to SARS-CoV-2 infection and COVID-19 susceptibility that do not yet have the same depth of clinical evidence as the top tier that the company also wants to stratify.
"Because the cost of the whole-genome sequences is a little bit of a challenge at this point, we are working with one of our partner labs to build a targeted gene panel that just looks at the genes that we're interested in," Raines said.
Breakthrough Genomics is not making the list of genes public just yet.
Raines said that the susceptibility risk assessments can help as US states seek to accelerate the reopening of their economies and gett people back to work even though the number of new COVID-19 cases continues to rise in most of the country. "We can contribute [by keeping] the people in isolation or quarantine that are likely to have adverse effects if they don't have the virus and if they would get infected," Raines said. "That, to me, is where we hopefully can add some value."
Raines said that Breakthrough Genomics is still struggling with assigning "weights" related to the significance of each gene. "The issue right now is we don't have enough data to actually generate some sort of algorithm for susceptibility. We have a set of genes and we have all these variants, but there's not a really great way to weight it," she said.
"I think right now we just want to partner with the clinical people and try to correlate genetic information with clinical outcomes," Li said. "Not every gene has equal weight, so want to be able to correlate with clinical symptoms and to assign the weight and to precisely predict the risk score."
While the process of understanding COVID-19 is still in the early stages, the company wanted to get the word out now that it could support accelerated studies and trials.
"In order to understand how our genetics plays a role in the viral infection and the viral response, it is critically important enough for the future for finding the cure or treatment," Li said. "To understand the picture on both sides, from the viral side and from our own genetic side will give us the full picture to understand the battle we're dealing with right now."
Breakthrough Genomics built its algorithms in house and released Enliter early last year. The company also offers sequencing services through an undisclosed lab partner.
Li noted that an explosion of sequencing as the price of genetic testing has come down has caused a bottleneck in understanding what test results mean.
"Our goal is to try to use the cutting edge of technology like machine learning and artificial intelligence to speed up the interpretation process," Li said. "We scan terabytes of genomic data and scan millions of published papers and give out meaningful interpretation to the users."
Raines said that Enliter is "lab-agnostic," so it can manage and analyze data from many sources.
The company has two patents pending for machine learning processes, related to how Enliter scans data and medical literature and provides interpretation, according to Li.
Breakthrough relies on cloud platforms rather than running its own high-performance computing center. The production version of Enliter relies on the Amazon Web Services cloud, while the company runs its R&D on the Google Cloud, Li said.