VastBiome is a next-generation drug discovery company that is mining the human gut microbiome for novel biomarkers and therapeutic molecules
Did you know that we have more bacteria in our gut microbiomes than we do cells in the entire human body? Or that the same gut bacteria produce 90% of our body's supply of serotonin (a neurotransmitter responsible for regulating mood in the brain)?
When you step back to think about how everything we ingest must first pass through and be processed by our gut microbiome (e.g. food, vitamins, pharmaceuticals, germs), it's easy to grasp the central role the gut is likely to be playing inside our bodies. And now, a fast-growing body of research is emerging that is directly linking the microbiome to the regulation of our metabolism, immune systems, and in determining how the body responds to drug therapeutics. As we continue to map out and understand the microbiome, of which 95% remains uncharacterized, there is a massive opportunity to unlock various new medicines and biomarkers.
VastBiome is at the bleeding edge of this exciting field. They are leveraging the power of machine learning, distributed supercomputing, and next-generation sequencing to disrupt the drug discovery process. The have developed computational and experimental methods to identify, prioritize, and isolate the key metabolites (bacterial byproducts) that are responsible for directing outcomes in the microbiome, which in turn represent starting points for drug therapeutics to drive specific outcomes in the body.
Imagine taking a naturally occurring metabolite that is effective in fighting disease or in guiding a positive response to a drug, and then building upon it to make it better and produce it in significantly greater volumes for patients who are lacking? Imagine a less than helpful metabolite that is preventing a drug from doing its job well or the body from effectively fighting off an invader such as COVID-19, and targeting it to either neutralize or reverse its behavior? That is what VastBiome is striving to accomplish and, in doing so, has caught the attention of the industry!
Today’s Status Quo
Pre-clinical challenges: Pharma companies often spend years of time and as much as $1B in the pre-clinical testing of a new drug only to end up with a 10% chance it being both safe and efficacious when finally tested in humans.
Post-market challenges: Pharma companies fortunate to make it to market with a blockbuster drug are heavily incentivized to understand why their drug, despite being extremely effective for some patients, does not work at all for others. For example, immune checkpoint inhibitors, a common drug therapy to fight cancer, work only 20% of the time on average. Understanding why that is and using this information to increase efficacy above this level has the potential to improve patient outcomes as well as make substantially more patients eligible for treatment, both extremely lucrative for the company.
Recent advancements in gut microbiome research has brought excitement and hope to the scientific and medical field, showing promise in the microbiome to be used as a modulatory component in one’s health that could drive better health outcomes. Building off of this body of research, Vastbiome asks, ‘Why build novel compounds from scratch when you can leverage known molecules in the human body already playing a part in patient responses to disease and drug treatments?’
Leveraging advances in next-generation sequencing, machine learning, and distributed supercomputing, VastBiome has developed a drug discovery platform capable of sourcing new therapeutic drug candidates from within the human gut.
They have developed computational and experimental methods to identify, prioritize, and isolate the key bacterial metabolites that drive these outcomes.
These isolated compounds provide excellent starting points for therapeutic development, as they are sourced directly from the highly contextualized setting of the gut of patients exhibiting the patient health outcome of interest.
This is a BIG deal.
Mapping out and characterizing the function of the gut microbiome is an opportunity to unlock various new medicines and biomarkers.
As the cost of bringing new drugs to market continues to rise which then is transferred to patients, biopharma is looking for new solutions to discover therapies. Natural products sourced from soil and plants use to be a high value approach. With all the drugs that have been developed we have tapped into only 1% of natural products to source therapies. Unfortunately the low-hanging fruit are gone and the field has shifted to rational drug design in the lab. This leaves a massive opportunity to investigate both soil/plants and the microbiome.
VastBiome focuses on the gut which has yet to be tapped into. While most research has focused on studying the human genome, roughly 30,000 genes, there are 100 times more genes, approximately 3KG of bacteria from the microbiome and yet we know nothing about them.
With a plethora of research demonstrating the connection of the gut microbiome to our health, we are at an inflection point in leveraging our bodies microbial properties to heal ourselves from disease. In order to truly capitalize on this opportunity, we need to understand what we are dealing with. That is where VastBiome comes in to support the characterization of the microbiome to drive better patient outcomes.
VastBiome’s Discovery Engine
VastBiome has developed a proprietary computational and experimental pipeline that identifies and validates key bacterial gene clusters and their respective metabolites that have an effect on the human immune system. This process provides a fully functioning pipeline to characterize the microbiome and its byproducts.
Mine the Microbiome. Analyze the gut microbiome of patients who have responded well to various diseases or to particular therapeutic treatments
VB’s bioinformatics pipeline featurizes the uncharacterized gene clusters from gut microbiome samples which they have collected and sequenced alongside their clinical research partners. Once featurized, VB maps out the gene clusters onto a graphical network to better understand which gene clusters are more alike than others. From there VB uses its AI and machine learning models to predict which gene clusters have an association with modulating the immune system. Once a list of prioritized genetic blueprints has been developed, VB runs in vitro experiments to validate these predictions.
Prioritize Bioactivity. Identify the active molecules produced by bacteria in the gut and rank which are most likely to be the drivers of therapeutic effect and the underlying patient response
VB’s in vitro assay is a novel approach to knockdown the gene clusters they have selected using a probe that they identify and synthesize. The probe selects key genes within the gene cluster to knockdown. If they are able to inhibit the gene cluster from expressing a phenotype of interest, that serves as confirmation that they are on the right path to identifying a novel biomarker or therapeutic lead.
Emulate Lead Compounds. Reverse engineer therapeutic compounds that mirror these integral compounds
Next, VB utilizes flow cytometry to identify whether the selected gene cluster does in fact have an effect on immune cells of interest. And if positive, they would run mass spectrometry work to identify which metabolites that particular genetic cluster produces. That would be the starting point of identifying a chemical structure with drug-like properties.
Develop Therapeutic Candidates. Validate the safety and efficacy of lead compounds through traditional preclinical testing
Another Layer Deep
The status quo in drug development today relies primarily on rational drug design in the lab and high throughput screening, with no context as to which chemicals may actually be viable drug candidates.
Biopharma is struggling with ways to find new therapies in a cost effective manner. Approaches including cell therapy and gene therapy show promise, but are complex and could lead to a list of new unforeseen adverse events downstream. In contrast, the microbiome is an opportunity to leverage our body’s own microbes to improve our health. Given that they are sourced from the gut, they are more likely to be safe and effective.
Within the Microbiome space itself, most companies are developing live biotherapeutics, which involve administering live bacteria as drugs. These methods have shown promise in a laboratory setting, however, a live biotherapeutic has yet to make it through clinical trials. The field is looking for that to change in the next 12 - 18 months. The challenge with live biotherapeutics is multi-fold: 1) the strength of the IP is considered weak, 2) manufacturing microbes at scale is challenging/expensive and 3) determining and standardizing proper dosage is more complex than a traditional small molecule drug approach.
When looking for bacteria that might have an effect, other companies are looking at a brute force approach, doing high throughput screenings of all molecules found in the gut. This is similar to traditional drug discovery, just focused on sourcing from the gut. The challenge here is like any drug discovery effort, a lot of time and energy is spent on sourcing through molecules without any context as to which ones are more likely to have therapeutic effect over the other.
Let’s contrast this with VastBiome’s targeted gene mining approach, which starts on the other end by first identifying patients that are clinically responding in a specific way and working backwards from there using advanced sequencing and computation to uncover which gene clusters and their respective metabolites (byproducts) are associated with their clinical response. This method allows VB to focus its energy on key signals of molecules that have a higher likelihood to be associated with influencing an immune response.
Separate from VB’s technical approach, the company’s business model is focused on being a drug discovery company, not a drug developer. The majority of companies, if not all in this field are focused on taking drugs all the way through human trials. VB focuses its energy on sourcing high-value therapeutic leads and working with biopharma to co-develop them. This gives them the opportunity to focus on building out the platform while working with drug development professionals to bring these products to market.
Technology. In building off of active compounds that are already present and effective in the human body, and can be tested and developed as traditional drugs, VastBiome’s disruptive approach utilizes proprietary bioinformatics and AI/ML models, and has the potential to substantially lower the cost, safety, and efficacy risks typically associated with new drug compounds.
Data Engine. The VastBiome data engine has amassed 13,000 patient gut microbiome samples. 5000 of which are whole genome shotgun sequenced. 1000+ proprietary gut microbiome and blood samples we sequenced ourselves. Moreover, VastBiome has taken the time to generate clean, tagged, organized, high-value data. Per the CEO of Novartis: “the first thing we’ve learned is the importance of having outstanding data to actually base your machine learning on. In our own shop, we’ve been working on a few big projects, and we’ve had to spend most of the time just cleaning the data sets before you can even run the algorithm. That’s taken us years just to clean the datasets. I think people underestimate how little clean data there is out there, and how hard it is to clean and link the data.” VastBiome’s focus on transparent model behavior, data featurization, robust annotation, and knowledge representation fit well with many crucial needs widely recognized by pharma partners, representing a valuable asset for partnership.
The company’s upcoming 1000-patient observational study will become the largest study in the world for patients with checkpoint therapy and testing their gut microbiome. This data is of great value to the biopharma industry broadly, with many companies having already approached VB to access their current and future data.
Team. Highly experienced, multi-disciplinary team with comprehensive expertise spanning immuno-oncology, the microbiome, clinical pathology, drug discovery and translation, genomics, computational biology, AI, and healthcare strategy (Management: Duke MBA, Hopkins/Harvard MD, Stanford Biomedical Data Science Dept/Harvard Big Data Systems Lab; Advisory: Mayo Clinic, Whole Biome, Second Genome, MD Anderson)
Leveraging Major Industry Trends
VastBiome’s technology approach is arriving at a prescient time for the industry, leveraged by four powerful technology waves:
The Rise and Success of Immune Checkpoint Inhibitors. ICI’s today have amassed $13B+ in annual sales and are anticipated to continue growing 12% annually. There are 150+ new ICI drugs in development, and 2250 ongoing trials today and growing, with some expanding the use of approved drugs into new disease areas and a majority being tested for use in combination with other drugs.
VastBiome is initially focused on non-small cell lung cancer which makes up roughly 60% of the entire ICI market. The team will be turning to the gut microbiome and its proprietary discovery engine to better understand why available therapies work incredibly well for 20% of patients in this segment and virtually not at all for the remaining 80%. It is believed that the information and therapeutic candidates emerging from this effort could be used to help determine which patients are most likely to respond to various treatments, to enhance patient outcomes for known responders to treatment, and potentially enable existing ICIs to have an effect on a higher percentage of patients.
The Rise and Success of Drug Repurposing & Combination Therapies. Pharma companies increasingly want to repurpose already approved drugs. For example, 95% of new ICI trials involve approved drugs. Pharma is also increasingly interested in combining individual treatments to fight cancer on multiple fronts, a proven approach to increase efficacy for patients. When successful, these paths represent new markets and revenue streams for pharma companies, creating substantial incentives to aggressively pursue each.
Attention on the Gut Microbiome. The field of microbiome research has made great strides in the past 5 years, with the gut microbiome starting as a mysterious, non-factor to now widely considered a significant factor in and viable channel for monitoring and improving human health. Going a step further, the gut microbiome has been identified as a factor that influences the body’s response to immune checkpoint inhibition therapy. This suggests the presence of bioactive substances in the gut with the capacity to serve as biomarkers for predicting patient responses and as cancer therapeutics to enhance the efficacy of checkpoint inhibitors. This data has fueled growing pharma industry interest in the space. In one such example, Merck, the owner of the blockbuster ICI drug Keytruda, is sponsoring a clinical trial wherein patients who previously did not respond to Keytruda are given a second course of the drug in conjunction with a fecal transplant (a means of transferring the entire gut microbiome) from a human who responded to the treatment.
Next Generation Sequencing & Machine Learning. Until recently, it was not possible to do what VastBiome aims to accomplish. Advancements in deep learning, distributed supercomputing, and next generation sequencing, and the significant drop in costs to perform these efforts have dramatically shifted the landscape, enabled and ripened the opportunity.
Initially, VastBiome is focused on is non-small-cell-lung cancer (NSCLC) patients who are being treated with immune checkpoint inhibitors. To date 60% of the ~$13B market for checkpoint therapy is in NSCLC with a 12% growth YoY. They aim to develop companion diagnostics (biomarkers that help determine how patients will respond to therapy) and combination therapies (combining multiple drugs to enhance patient efficacy) for various checkpoint therapies. There are over 2250 ongoing trials today for checkpoint therapies, and growing. Each of these is in a prospective customer/partner for VastBiome’s work.
In the future, the company plans to expand to therapeutic areas outside of Immuno-oncology, including gastroenterology (e.g. IBD $7B) and autoimmune diseases (e.g. Rheumatoid $19B). The $24B medical foods industry is also one of keen interest to VB considering the microbiome’s role in processing foods and regulating metabolism.
MBA (Healthcare Management) | Duke University
Business Operations & Sales | Google
Co-Founder | StingyCampus
Kareem Barghouti is the co-founder and CEO of VastBiome. He received his B.S. in Biological Sciences at UC Irvine and a MBA from Duke University with a concentration in Healthcare Sector Management. His background is in Strategy & Operations and Sales & Marketing. Kareem joined the fast growing startup Wildfire Interactive as a Business Development rep which was acquired by Google for $450M. He spent 5 years at Google leading the data integrity of Google’s advertising business, managing teams of 100+, and was part of the core product team to launch an online education portal to millions of users globally. He was recognized as the top sales rep globally (1000+ reps) for the New Business Sales team. Before Google, he co-founded an online marketplace (StingyCampus.com), having raised $250K in seed funding and launched the program across 12 universities in Southern California. Most recently, Kareem participated as a Biodesign Fellow at the Texas Medical Center Innovation Institute in Houston, Texas, where VastBiome was born.
MD | Johns Hopkins University
Pathology Chief Resident | Harvard University
Co-Founder | Accetia (Microbial Bioinformatics)
Peter McCaffrey is the co-founder and CTO of VastBiome. He is a Board-Certified Pathologist who received his medical degree from The Johns Hopkins University School of Medicine and completed his Clinical Pathology Residency at The Massachusetts General Hospital/Harvard Medical School where he was also Chief Resident. In addition to his medical experience, Peter also has experience and skill in Data Science and Bioinformatics. Prior to Residency, Peter was a Research Fellow in South Africa where he built and managed a Genomics Core facility including selection and installation of sequencing machinery, development of SOPs and fee schedules, and the development of robust bioinformatics capacity including the brokerage of a $250,000 grant from IBM Research for server infrastructure to support custom bioinformatics pipelines. Based upon this work, Peter founded Accetia which built genome analysis pipelines for the purpose of profiling antimicrobial resistance in tuberculosis and enabling personalized therapy. At its peak, he managed nine engineers and built a comprehensive computational biology platform which included a web application with an elastic cloud backend for analysis, a plugin that was installed onto physical sequencing machinery, and an iOS application that served reports. Peter has a strong background in Python and R, with a broad and deep skill set in Bioinformatics. Moreover, he has a deep understanding of the therapeutic applications and limitations both of immunotherapy and of molecular profiling in cancer. Most recently, Peter participated as a Biodesign Fellow at the Texas Medical Center Innovation Institute in Houston, Texas, where VastBiome was born.
Microbiology & Drug Discovery | University Of Vermont
General Surgery Resident | Stanford University
Research Fellow | Centers for Disease Control
Kovi Bessoff is the co-founder and CSO of VastBiome. He completed his MD/PhD at the University of Vermont. As a surgeon-scientist, Kovi has extensive experience in clinical medicine and microbiology and brings these perspectives to VastBiome to help hone the solution and value proposition to be both scientifically feasible and valuable. He is well versed in the process of identifying and validating molecular leads. Kovi’s PhD dissertation focused on the discovery of novel therapies for cryptosporidiosis, an intestinal infection caused by waterborne parasites from the genus Cryptosporidium. His work led to the development of the first high throughput cell based assay (recently adopted by industry), and therapeutic leads (currently being optimized in animal models) for less than $100,000. In addition, Kovi had spent time at the Centers for Disease Control in San Juan, Puerto Rico and was in charge of coordinating an international effort to standardize dengue virus diagnostics. He built a database and a serum library of diagnostic standards which drastically increased the efficiency of the project. As a second year general surgery resident, he developed a consult handbook and identified opportunities for the general surgery department to use it in conjunction with other tools in the literature to create a framework for outcomes based research. The system is still in use and will provide the foundation for a burgeoning clinical research program at Stanford. Most recently, Kovi participated as a Biodesign Fellow at the Texas Medical Center Innovation Institute in Houston, Texas, where VastBiome was born.
Chief Data Scientist | Foundation for Precision Medicine
Fellow | Harvard Launch Lab X
M.S. Information Systems | Harvard University
Ayin Vala is the co-founder and Chief Data Science Officer of VastBiome. He has over eight years of experience in predictive analytics and machine learning. He is an instructor at Stanford’s Biomedical Data Science Department and is the Chief Data Scientist at the nonprofit Foundation for Precision Medicine. His team works on statistical analysis and machine learning, pharmacogenetics, and molecular medicine. On this effort, he started partnerships with Stanford University, Mayo Clinic, Yale School of Medicine, and Google. He previously developed a deep learning algorithm to detect lung cancer in at risk patients, collaborating on a clinical trial at The Johns Hopkins University School of Medicine. He was a researcher and active resident at the Big Data Systems Laboratory at Harvard University School of Engineering, and at the Launch Laboratory and Innovation Laboratory at Harvard Business School. Ayin is the leading inventor of several patents and recipient of multiple awards relating to the application of machine learning in Healthcare, Biotechnology, and Aerospace Engineering, including a patent approved by NASA. He holds a Masters degree in Information Management Systems from Harvard University.
Role of the Gut Microbiome in Immunotherapy
Machine Learning, Neural Networks, and Inference
Axes of Microbiome-Based Immunomodulation
Microbial Genome Mining