Engineering Biology: Announcing Fresnel

Picking the Right Targets
The hardest problem in biopharma today is picking the right targets. Our ability to modify biology has increased exponentially over the past decades. No longer is it a question of if we can hit a target (or a pathway) with some compound. Today, we can hit nearly any biological target with multiple different modalities from traditional small molecules to antibodies to interfering RNAs to cell and gene therapies and beyond. The key questions today are what should we hit and how.

Despite these technological advancements, drug discovery is still fraught with exceptionally high failure rates with up to 90% of drugs failing in the clinic. Poor choice of biological targets drives these failures with toxicity and lack of translational efficacy being especially common causes. Targets are too often chosen by “gut feel” or “expert opinion” leading to industry-wide herding into a handful of known targets and ignoring discoveries from the last two decades of biological discovery.

This is not for lack of evidence about targets. 20+ years after the Human Genome Project, we have a wealth of genomic evidence about genetic targets, their connection with phenotype, their expression patterns, their link to drug safety, and beyond. Genomic evidence is associated with considerable increases in program success (~3x¹). Notable cases include PCSK9 (cholesterol), TYK2 (autoimmune disease), and TL1A (GI disorders). Yet this information is surprisingly rarely used. A recent review² has shown that despite demonstrable successes, only ⅓ of active drug programs use genomic evidence.

Despite ample public genomic repositories, many of these datasets are obscure, isolated, and not easily accessible. They live in disparate databases, are curated and analyzed with uneven quality, and are difficult for scientists, entrepreneurs, and investors to access and use. With the right computational biological knowledge and technical support, they could be made much for impactful.

Extant tools and data summaries are unimodal: GWAS catalogs, databases of transcriptomic experiments, maps of biological pathways, curations of clinical trials. Each of these lines of evidence is interesting on its own, but rarely dispositive. By aggregating these data modalities and allowing them to be viewed in the context of one another, dramatically greater insight can be gained, driving better decisions both about what to work on and how.

At Fresnel, we are building intuitive and beautiful software products enabling founders, executives, scientists, and investors to understand the association of a genomic target with disease, its biological context, the feasibility of drugging it, its potential safety risks, and the competitive landscape of clinical trials related to it. By making these data easily accessible with clear provenance and traceable analytics, these products can be used confidently by a variety of decision-makers without them needing to hire and manage consultants or computational biologists.

With an integrated view, users can see the association of variants in a gene with disease, the cell types in which that gene is expressed, whether there are connections between expression level and disease, whether other lesions in that gene lead to unexpected, off-target phenotypes, what biological cascade(s) it is part of and how to affect them, alongside what happened when drugs against that gene were trialed in the past. Such a view gives clear opportunities not only for assessment of the causality of the target in disease, but also what key experiments should be performed and a sense of the commercial opportunity.

Unlike too much software in biopharma today, Fresnel is built on modern software engineering principles and cloud computation infrastructure. Cost-effective, ubiquitous storage and compute enables a serverless architecture to power web applications serving computational biology analytics. This includes up-to-date reference genomic datasets (e.g., biobanks and single cell expression atlases), continually fresh analytics using state-of-the-art methodologies, traceability and auditability of all analytic results, and seamless, excellent user experience through modern web application design.

At Fresnel, we are inspired by the rich history of information products transforming growing industries. In the 19th century, a continual bestseller was Bowditch’s American Practical Navigator, which taught the elements of navigation to generations of seafaring merchants. In the 20th century, Bloomberg revolutionized finance through continually fresh, critical information delivered electronically. Our goal is to build the “Bloomberg of Biology,” giving everyone the tools of an expert computational biologist, revealing key insights for drug development and target validation from the vastness and complexity of genomics.

Curious to learn more? Reach out to us at hello@fresnel.bio


Illuminating the Known
In building Fresnel and illuminating biology, we were inspired by the eponymous physicist and optics pioneer, who illuminated the 19th century. Modern lenses are a product of the Renaissance with the telescope and microscope dating to the 17th century. Improvements in engineering led to the development of optics as a science and a clear sense of how to make light behave for practical applications.  The problem with lenses, though, is that the ideal spherical lens becomes hopelessly large as the area to be illuminated grows. This was especially true for lighthouses, leading to an active research effort in Napoleonic France, solved by Augustin-Jean Fresnel.
‍

‍
Fresnel’s insight was that a lighthouse lens needed only to illuminate. Light from a single source  would produce rays at all angles: each of these would pass through a small piece of lens that would refract it to fit a single, outgoing beam. In a lighthouse, there was no need to focus incoming light from the outside, which would be helplessly chopped up by the design of Fresnel’s lens.

Fresnel took a problem too complicated to be scaled and applied generally (the ideal lens) and simplified it to one that could be mass produced (the Fresnel lens). We at Fresnel are taking the same approach to the complexity of biological information. A lighthouse illuminates the seascape, enabling better navigation. A natural, simplified series of views illuminates the biological landscape, enabling the drug developer, the executive, or the investor to navigate better and to focus their attention on what matters.

Taken as a whole, each data modality is deeply complex: genotype phenotype relationships can be interrogated at the level of the gene, the variant, within-family associations, correlations within populations, and so on. The very breadth of available information perversely obscures the underlying signal. With easy-to-use, accurate representations at a higher level, though, users can pick and choose how and where to go deeper (or not to). Fresnel Bio illuminates the landscape of biology.



¹ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7614359/
² ibid