NEW YORK (GenomeWeb) – With a number of planned upgrades to its Genome Identification Universal System (GENIUS) metagenomics analysis software and database, CosmosID is trying to position itself as a provider of a one-stop digital assay for detecting and characterizing harmful pathogens in both research and clinical contexts.
The company has raised $6 million in a Series B funding round led by private investment firm Applied Value Group, which it is using to hire new staff and to support clinical applications of its platform, as well as to develop new products for monitoring, predicting, and preventing infectious disease outbreaks. The company will also fund research into methods of making better genotype-to-phenotype correlations and scale up its computational platform to better support more users and handle larger quantities of sequencing data.
CosmosID offers desktop and appliance-based options of its GENIUS software as well as cloud-based applications on Illumina's BaseSpace that provide capabilities in the company's broader system. These tools offer access to algorithms for analyzing metagenomics data and curated databases of more than 65,000 microbial genomes including bacteria, viruses, fungi, parasites, and antibiotic resistance and pathogenicity markers. GENIUS, its developers claim, can identify difficult to detect microbes, such as those associated with chronic wounds, respiratory infections, meningitis, and encephalitis, as well as pathogens involved in co-infection and polymicrobial infection cases and multi-drug resistant organisms.
The firm claims that given a single input sample, GENIUS can detect all pathogens present as well as provide information about their virulence and any resistance genes, and return results in less than 24 hours. The software also calculates a statistical score that provides a measure of confidence in the accuracy of the identifications that the system makes, a feature that the firm said sets it apart from competing metagenomics analysis firms.
"The system that we've developed is very user friendly ... you don't have to be a highly-trained bioinformatician to be able to enter your data and get results," Rita Colwell, a professor at the University of Maryland, College Park and CosmosID's founder and chair, said in an interview with GenomeWeb. Also important "is the high accuracy of [its] identification and the fact that ... this is really tailored for understanding microbial communities in general."
The privately-held company, which began generating revenue in 2015, has secured contracts with the US Food and Drug Administration, where researchers are using the platform to identify pathogens associated with foodborne illness outbreaks. Other GENIUS customers come from the pharmaceutical, probiotics, and microbiome research spaces.
The company is now trying to push its platform into the clinical space. It has already started, and in some cases, completed a number of pilot studies, and it has plans to run additional validation studies with a total of 15 research hospitals in the United States and Europe. "Right now, we are focusing on clinical research [use cases] to demonstrate the clinical utility of this universal assay to detect a broad range pathogens and to distinguish infection in these same pathogens from the commensal flora associated with different biological samples," Nur Hasan, CosmosID's vice president, told GenomeWeb. "Also, we are fleshing out the ability to differentially detect and distinguish infection-inducing pathogens from background contamination."
They also plan to carry out large-scale studies "where traditional resistance phenotyping will be used in conjunction with predictive genotyping," he told GenomeWeb. Having the ability to link genotype and phenotype is an important feature for clinicians, Colwell noted. "You are not only using this to identify the strains ... [and] identify the presence of antibiotic-resistance genes, pathogenicity traits, etc.," but also to answer the question of whether they are functioning, she said. That translation of genotype to phenotype is "absolutely an active direction in which we are going."
Colwell and colleagues at the University of Texas Medical Branch, led by Ashok Chopra, a UTMB professor of microbiology and immunology and lead author, recently published a paper in the Proceedings of the National Academy of Sciences that described interactions between multiple strains of the bacterium Aeromonas hydrophila in a necrotizing fasciitis — flesh-eating disease — case. In the study, the researchers discovered that although there were four strains of A. hydrophila present in samples collected from the patient, only one of the strains was responsible for the infection.
Chopra's team knew from a previous study that three of the strains were similar to each other, while the fourth, although from the same family, was phylogenetically different. They assumed that all four strains were responsible for the infection observed in the patient. However, when they used GENIUS to analyze data from liver and spleen samples from infected mouse models, they found that only the phylogenetically different strain was present. Follow-up testing in more mice, using the luciferase gene to mark the two kinds of strains, confirmed that only a single strain — labeled NF-1 in paper — was responsible for the infection.
The researchers further discovered that the three similar strains produce a toxin that breaks down muscle tissue, according to the PNAS paper. Once the tissue was destroyed, the NF-1 strain — which does not make the muscle-destroying toxin — killed off the competing strains and proceeded to infect the host organism's liver and spleen.
Discoveries like this could alter the way medical researchers think about bacterial diseases that are commonly thought to be caused by a single species of bacterium and demonstrate the value of a metagenomics approach to infectious disease analysis.
"Polymicrobial infections are a very hot topic," Chopra told GenomeWeb. "There could be a synergistic effect of such infections on the host in terms of the severity of the disease, [meaning] it could be more serious than if those individual pathogens were causing infection by themselves." The results of their study up the ante further because they show that an infection caused by a single species could also be a poly-microbial infection, information that could have implications for treatment options.
For example, "it is quite possible that these different strains of A. hydrophila may have sensitivity to different antibiotics," he said. Depending on when and where the samples are taken from, a different strain might be more dominant in cultures than in the patient. So while a particular antibiotic might prove toxic to the cultures, it might not be quite so quite effective when it's given to the patient.
Using ComsosID's approach to figure out which strains are actually present in the sample, a clinician can come up with a more appropriate treatment strategy, for example, using a combination of an antibiotic and antibodies to more effectively combat the infection, he said. "That's why we are excited about this [approach]. We are actually putting together a grant on this with Dr. Colwell's group ... and designing different experiments that we can do," said Chopra.
Other clinical studies that the company has been involved in included a collaboration with a University of Maryland School of Medicine group led by Rose Marie Viscardi, a professor of pediatrics in the neonatology department at the university. The project involved analyzing samples from newborn babies who might have been exposed to some kind of infection during birth.
In cases where clinicians suspect a newborn might have some sort of infection, they collect a blood sample, culture the microbes present, and then do standard antibiotic susceptibility testing to identify the source of the infection. But it takes days to get answers and the results are often negative because the original samples sizes are so small to begin with and very few bacteria actually grow in the cultures, Viscardi said. "We have a baby who might have an infection, who we can't prove doesn't have an infection. So, sometimes we end up treating with antibiotics anyways," she said.
The result is longer hospital stays for the mother and baby, and a significant bump in healthcare costs. According to Viscardi, some 250,000 babies born each year will go through this process of being evaluated for infection and treatment with antibiotics to the tune of about $68 million a year. The actual rate of infection, however, she said, is only one to five per thousand live births, meaning more babies are being treated than need to be.
"The beauty of the CosmosID strategy is that it could rapidly, with very small amounts of blood — as little as one hundredth of a milliliter — detect bacteria," she told GenomeWeb. "And they have already put together a database of the most common bacteria, viruses, and fungal infections that can cause infections in babies, and so they can pretty rapidly identify the possible infections." Furthermore, the company's platform is far more sensitive than cultures, she said, adding that "they can identify when it's truly negative, so that we could then safely either not start antibiotics or be able to discontinue them so that the baby doesn't have to stay longer in the hospital and be exposed to a treatment they don't need." Also, clinicians can simultaneously test for the presence of bacteria, viruses, fungi, and other microbes in a single sample, as well as figure out which treatments these microbes might be sensitive to.
Viscardi believes that with further testing and validation, a system like GENIUS could ultimately replace standard microbial testing methods. "We would have to show that it identifies the same bacteria or virus as the culture or other testing techniques might identify," she said. "We have planned a study to evaluate ... this as a diagnostic tool specifically for neonatal infections."
A third study, still in its infancy, is being done in collaboration with researchers at the University of Wisconsin-Madison and Iowa State University. It's a National Institutes of Health-funded study using monkey models that is exploring how babies acquire their gut microbiomes.
"The idea is that infants are mostly born sterile and acquire their first gut bacteria from their mother, and there's a lot of interest in that with respect to natural delivery versus C-section [and] breastfeeding," Christopher Coe, a professor in UW's department of psychology and one of the PIs on the study, told GenomeWeb. There's also growing interest in infants' gut microbiome because of evidence that indicates that "the gut microbiome primes the development of the immune system [and] predisposes or not to allergies or tolerance, and there's evidence that it affects brain development," he added.
For now, the researchers are analyzing data from stool samples, although they will eventually look at samples from the test subjects' large and small intestines. They provided CosmosID with data from about 50 stool samples, which they'd also analyzed using 454 and 16s Illumina sequencing — they plan to compare the data from those analyses to CosmosID's results.
The early results from the GENIUS platform are very promising, according to Coe. "One of the things I was impressed with was the turnaround time," he told GenomeWeb. "We first sent specimens to them in late October, and we are already talking to them about results, so for me that was impressive that we were able to get such fast turnaround." Coe also noted the system's ability to provide results at the species level rather than the genus level. "I was impressed by the number of species that were identified, particularly ones in lower abundance," he added.
CosmosID CEO Manoj Dadlani said the firm is also working out with its clinical partners how best to report results. Rather than simply return a comprehensive and cluttered list of all the genomes found in their samples, "it's important that we can deliver a report that's clinically relevant to a physician," he told GenomeWeb. "So, that's something that we are actively working on as well."
In addition, the company is increasing the capacity of its computational platform so that it can cope with even larger quantities of sequence data. Dadlani said that the company has hired four more developers — it now has a total of seven on staff — to build the computational architecture it needs. "There's a huge amount of sequencing that's going on and in order to continuously maintain the databases we need to have, we need to think differently in terms of how we do our development and our curation." CosmosID also has a number of partnerships in place that will increase the number of customers using its platform, and it needs to cope with growing demand from customers of its BaseSpace application. "We are building infrastructure so that we can handle definitely thousands of users and that wouldn't be a problem," he said.
The company will also develop tailored solutions that address specific needs in the different market segments that it targets, Hasan said. Clients will also be able to use the company's cloud application in different ways, as well as customize their appliances — for those who have them — to support high-throughput applications or to meet privacy requirements, for example, he told GenomeWeb.
CosmosID also has signed partnerships with a number of yet-to-be-named genomics companies, Dadlani said, which the company will announce later in the quarter. It will also provide updated pricing details later on this year.
The company operates under a per-sample pricing model, but it also offers volume discounts and special pricing modalities for clinical versus non-clinical users and for academic and commercial users. Previously, it charged between $200 and $550 per sample. Dadlani said that the new price point would be on the lower end of that range.
The company also plans to update its BaseSpace offering to include both free and paid versions of the GENIUS app.