Axial Podcast

Melih Yilmaz is a PhD student at the University of Washington. Where he focuses on computational biology and develop machine learning tools. Advised by William Noble and Sewoong Oh, Melih's current research interests are around proteomics, particularly building deep learning methods to analyze mass spectrometry data. We start off the conversation talking about his journey to biology and consequently the United States. Around 2016-2017, deep learning had began to gain pace, drawing Melih in. Around the same time, peers of his were going to the US. Slightly influencing Melih to land a research internship at Stanford with Tina Hernandez in the biomedical informatics department. He then entered a PhD program at Washington after several other internships. We go into various experiences of Melih's as an engineer in biology. Expanding on his work in protein sequencing.

Peter Turnbaugh is a professor at UCSF studying the human microbiome’s effect on pharmacology and nutrition. In our conversation, we discuss his journey to become a scientist and help pioneer the microbiome field starting in graduate school. We talk about his research as a fellow at Harvard then professor at UCSF, and his lab’s current work. A key theme across the conversation is learning by doing.Peter’s work has been anchored around predicting and controlling the metabolism of complex microbial communities. Going to a liberal arts college, Whitman College, Peter gravitated more to science because the grades seemed more objective. An experience in a stem cell lab helped Peter learn that science is not just a series of facts and helped spark him to apply to a PhD program. He applied to a bunch of schools, getting rejected by many of them like UCSF, and got into 2 programs. Peter ultimately chose to attend WUSTL. This might have been one of the best, and luckiest, decisions Peter made. Washington University in St. Louis was an epicenter for the Human Genome Project and is a hotbed of genomics/bioinformatics talent. Going in his rotations, Peter was excited about genomics and wanted to work on more computational problems. He ended up joining Jeffrey Gordon’s lab, a legend in genomics. Jeff gave Peter the nickname, “Professor,” almost foretelling his future. Ruth Ley had also just joined the lab as a postdoctoral fellow after working with Norman Pace and wanted to study microbes across mammals. She had gone to the St. Louis Zoo to collect samples. On Peter’s first day in the lab, the fridge was full of wild animal poop to study. At the time metagenomics was not widely accepted - many genome scientists thought microbial communities would be impossible to understand. This is around 2004 and there wasn’t a checklist to follow. Peter, Ruth, and the lab had to not be shy to ask for help in order to pioneer a new field of the human microbiome. This ultimately led to 2 papers in 2006 that established gut microbes associated with obesity.After writing up his thesis, Peter wanted to “keep going as far as he can go” with science. He became a Bauer Fellow at Harvard, rejected by UCSF again during this process, and was excited to start his own lab. Similar to his grad school research, Peter learned by doing to build a lab. The lab’s initial idea was to switch from studying host-associated microbes' role in diet to pharmacology. Xenobiotics seemed “really weird” and an open field to study. A 2013 paper showed how diet can alter the human microbiome, where the lab had to run the trial themselves - feeding participants a veggie or meat-only diet over 5 days. This type of work helped Peter finally get into UCSF and become a professor.The next skill he had to learn while doing was grant writing. Up until then, Peter had only written grad school applications and a 3-page proposal to Harvard that gave him 5 years of guaranteed funding. At UCSF, he’s built a highly successful group leading the way on the study of the human microbiome and how it intersects with diet & pharmacology. He’s expanded to work on CRISPR/phage research now, and excited to see how that thread grows over the next few years. Looking back, Peter has gained a greater appreciation for the community, something his grad school mentor, Jeffrey Gordon had emphasized. Just as much as multiple genes affect something like height or a disease, it’s important to go beyond a single microbe causing a phenotype.

In our conversation with Wyatt McDonnell, the Co-Founder and CEO of Infinimmune, we discuss his journey to 10X Genomics, his work there, and the founding of Infinimmune. Wyatt is a world-class inventor & immunologist working across a wide range of projects at 10X from launching BEAM (barcode-enabled antigen mapping), working on the Immune Profiling v2 product, and developing various immune repertoire technologies. After making a significant impact on 10X, Wyatt along with 4 other colleagues that had worked together founded Infinimmune to transform human antibodies into drugs. A key theme of the conversation is that tool users greatly outnumber tool builders. This creates an opportunity for the latter to build applications before others. Infinimmune is hiring across business, engineering, and biology roles. And is a home for tool builders. Please get in touch via founders@infinimmune.com if you're interested in applying your expertise to the next generation of antibody therapeutics. Wyatt got his first exposure to science from a Breaking Bad consultant that showed him a redox reaction. To Wyatt, it was like “watching alchemy unfold.” He went to Hillsdale, a small college in Michigan where the largest science class had 60 people. This intimacy played a key role in Wyatt’s development and allowed him to speed up his education. He thrived in organic chemistry as a freshman and worked with Frank Steiner, taking 4 of his classes and working in his lab where Wyatt caught the research bug. He had been thinking about medical school but decided to go to Vanderbilt for a PhD program focusing on immunology & precision genomics. While in graduate school, Wyatt saw 10X Genomics’ Chromium products taking off. People were coming back to use their instruments, something that is pretty rare in scientific research. This motivated him to apply to 10X and end up joining them to build solutions on top of Chromium. Coming on board in the summer of 2019, the first 90 days were spent figuring things out. Wyatt worked on a wide range of projects, and a key reason was his immunology expertise. He had a strong knowledgebase on TCRs, which was pretty rare at 10X at the time, allowing Wyatt to work across divisions. For example, he got to work with David Jaffe, a co-founder of Infinimmune, on the Enclone project. For Wyatt, working at a market leader like 10X was the perfect example to follow when he built his own startup.After leaving 10X Genomics in the summer of 2022, Wyatt started Infinimmune with people he had worked with at 10X. The company is built with a strong conviction around 3 pillars: We still don’t understand immunology that well Adoption of immunology tools has been too slow The best antibodies come from humans The platform is centered on analyzing healthy & disease samples to uncover rare, human antibodies. This requires deep analytical expertise, at the raw sequence read level, to maximize the value from sequencing samples. Infinimmume is building the toolkit to discover effective antibodies from patients that might have survived cancer or have protective genes against Alzheimer’s. On this point, for Wyatt, “platforms are only as good as the assets they produce.” Wyatt and the founding team are setting the foundation of the company right now to scale up partnerships and drug development over the next 2 years. Excited to get the update from Wyatt then.

Margaux Pinney is a Sandler Fellow at UCSF and in our conversation we discuss her journey to become a scientist & leader in her field and her work around enzyme evolution. The Pinney Lab studies how enzymes work, how they got there, and how they will adapt in the future. Margaux grew up in a small town outside of Seattle; Black Diamond, WA named after the high quality coal the town used to produce. As someone "obsessed with details," she thrived in chemistry while in high school. Margaux first considered chemistry as a potential career after a teacher told her she was good at chemistry, Margaux then entered community college, leaving her senior year of high school. Taking organic chemistry and physics, she knew she would pursue science and with her community college professor Keith Clay as a major supporter, Margaux entered the University of Washington for college, working in James Mayer’s lab. Where she began her research on enzymes. And coming from chemistry, to Margaux, enzymes are pretty much magical. Near the end of college, Margaux wrote an NSF proposal around directed evolution, and after getting the fellowship, she went off to graduate school at Stanford. In 2014, Margaux joined Daniel Herschlag's lab. But also did a master's in medicine, concurrently taking the first 2 years of medical school (the hardest 2 years by the way). To her, the master's program felt like an opportunity because she didn't have to pay the medical school tuition. A common theme in her career: taking full advantage of experiences. While at Stanford, she wanted to study enzymes. After a few references and a gut feeling she chose Daniel's lab. In retrospect, Margaux found the best mentors are committed, push you, and have the time. She initially worked on the RNA side. Telling a story she now uses as a proverb in her lab on doing one experiment at a time - Margaux, early-on, ran 9 gels at once, standing a foot tall and wide. Every single gel failed, and she learned to stick to one gel at a time. Margaux also learned she didn't want to work on RNA and moved to protein enzymes. She began studying the role of hydrogen bonds in enzyme function. "Enzymes have a history" and are actively evolving. Margaux wanted to figure out the rules that drive enzyme adaptation. Starting as a side project, she began studying temperature adaptation. Something Daniel said not to work on but Margaux did anyway (Daniel and Margaux now joke about this). Ketosteroid isomerase (KSI) is a model enzyme system with about 70 years of research. But with just 2 orthologs studied out of 1000s, its evolution was overlooked. By shining a light on old problems with new tools, mainly a large database of enzyme sequences generated over the last ~20 years, Margaux was able to discover a pattern of single amino acid substitutions that are stabilizing at higher temperatures. Publishing her work in Science in 2021. Throughout her journey, Margaux has learned that a high degree of commitment and flexibility are essential in science because it's both cyclical & a marathon. Around the time when the COVID-19 pandemic began, she started writing her thesis and won a fellowship at UCSF to start her own lab. She deferred to join the labs of Polly Fordyce & Gavin Sherlock as a postdoctoral fellow, learning their high-throughput methods for enzyme screening. In January 2022, Margaux got her lab off-the-ground to combine high-throughput screening and evolutionary analysis, increasing the throughput of enzymology. By bringing quantitative, high-throughput assays to enzymes, the Pinney Lab investigates enzyme function at a scale beyond traditional biochemistry. Work here is just getting started and excited to see the work Margaux & her group put out of the next few years.

Excited to put out a conversation with Devon Cayer, Co-Founder and CEO of 1859, a platform biotech company merging pico-scale screening and AI to scale small molecule discovery. We talk about his story that led him to founding 1859, how the company was built, and the long-term vision. Devon discusses how to set and measure core metrics for drug discovery platforms and business model design. A key theme across the conversation is "[creating] an ecosystem of solutions to problems." A framework Devon picked up during his scientific career and still applies to growing 1859. Devon first got interested in science through Discover Magazine and Jurassic Park. Then fell in love with chemistry while in college at UCLA. He then went to grad school at Scripps working under M. Reza Ghadiri on everything ranging from molecule computation to DNA barcodes and enzyme therapeutics. The lab was actually nicknamed the "Wizards and Warlocks" lab due to the seemingly crazy ideas they pursued. While there, Devon had several formative experiences around his interests in startups and solving problems in medicine. After his PhD, he went on to co-found MYi Diagnostics, learning a lot of lessons along the way. He then went on to work at Singular Genomics and Omniome to learn more about scaling a business. While at a Scripps seminar by Richard Lerner on DNA-encoded libraries (DEL), Devon saw an important problem: DELs were relegated to binding-based assays and were not functional or cell-based. Knowing he could screen large libraries at pico-scale and barcode them from his training at Scripps and work at Singular/Omniome, Devon sensed he had a solution for DELs. After assembling the founding team through frequent lunches and talks, 1859 got started in 2018. Named after the publication date of On the Origin of Species by Charles Darwin, 1859 generates functional data for large libraries of chemical matter. The founders began pitching and refining their platform - Devon talks about fundraising and team building here. They have been able to build a platform that can screen millions of compounds every week. 1859's workflow is made up of 4 parts: (1) A target, (2) Tailor-made libraries of 100Ks of compounds, (3) Pico-scale screening to create a structure-activity relationship for each compound, and (4) Using machine learning to identify the best leads. During this part of the conversation we touch on core metrics of the platform and how it was built piece-by-piece. After talking about the technology, we move into the business model of 1859. Inspired by AbCellera, Adimab, among others to spread out risk of capital intensity, 1859 is a catalyst for drug discovery. The company has built an ecosystem for small molecule discovery with multiple ways to plug into their platform: a Discovery Access program where partners gain access to workflows & libraries, Technology Access where 1859 builds out their technology and capabilities at a partner site, and Program Access where 1859 identifies their own Targets or works with academic institutions to develop medicines for their own pipeline. The company is just getting started but Devon and the entire team have done an impressive job to an idea originally inspired by a scientific talk into reality and a product touching a large set of drug developers and partners.

In our conversation with Abhishek Jha, Co-Founder and CEO of Elucidata, we talk about what it takes to build a SaaS company in biotech, the impact of AI on life sciences, & the prerequisites to build a startup. Abhishek and I discuss his journey as an academic scientist to Agios Pharmaceuticals to founding Elucidata in 2015. From his journey, a key quote is that “life doesn’t let you do a control experiment.” While making the transition from academia to industry, Abhishek was also exploring D.E. Shaw, where he would have probably stayed to this day, and Elucidata might have never been started. While at Agios, Abhishek was enabling the entire organization with data and spent a lot of time cleaning up data for drug developers. Elucidata was founded as an extension of this work. Data scientists often spend 80% of their time cleaning up data and 20% doing analysis. Elucidata’s mission is to flip that and let their customers spend more time on analysis. Elucidata’s focus was as a product company from day one. We talk about what it takes to build a software company in biotech, given the constraints on the number of customers, and the value of services in the beginning to find product-market fit (PMF). Agios was Elucidata’s first customer, with customers funding product development in the early days. Bootstrapping to find PMF forced Elucidata to make products customers will pay for. After raising a friends & family round, Elucidata’s growth began to pick up where the company works with 4 out of the top 10 biopharma companies and is the market leader for structuring genomic data particularly RNA-seq (bulk and single-cell) with emerging products in proteomics and other -omics. Their products focus on NLP and data engineering to standardize data sets across life sciences. A comparable company is Fivetran, a data aggregator for enterprises. For SaaS in biotech, there is a balance between building versus buying, and as a startup developing complementary products without initially competing against your customers. At the end of the conversation, we talk about the need for a strong support structure behind you to found a successful startup. For Abhishek, startups are “one of the best self-improvement processes.” And Elucidata is just getting started. For example, Fivetran took a few years beyond hypergrowth due to the technical difficulties of structuring large, diverse datasets. Following that trajectory, Elucidata is just getting started.

In our conversation, Bryan discusses everything from directed evolution and drugging RNA to what it takes to start a lab. The Dickinson Lab at the University of Chicago is a unique group composed of biochemists & synthetic chemists to cell biologists & synthetic biologists. The lab set up shop in 2014 to use chemistry to control biology with both evolutionary and rational methods. Bryan's research is heavily influenced by his career starting as an undergrad at Maryland with mentorship from David Fushman to graduate research with Chris Chang at UC Berkeley and a postdoc with David Liu at Harvard. At each point, picking up new tools to work with and models to follow. Bryan "works with people he is inspired by." And that has been a key driver of his group's success. To run such an interdisciplinary lab, Bryan focuses on important problems rather than a particular technique or tool. This not only cultivates openness to new solutions but aligns everyone around a shared passion & purpose. Bryan's lab has 3 main areas of research - (1) Chemical biology for protein lipidation, (2) Biosensors to control PPIs, and the (3) Epitranscriptome. We touch upon several examples here, but the key theme is inventing new, functional molecules, whether they are small molecules, proteins, or engineered organisms, to make breakthroughs in biology. This molecule-agnostic approach requires the ability to synthesize small molecules, screen large libraries of molecules, and use directed evolution to reprogram cells and design proteins. Finally, we discuss how to determine what is a valuable problem to work on. A skill that is nurtured with support from mentors and a team. A common thread across the conversation is the power of evolution, nature's way of designing things, to not only optimize but uncover new forms of biology. Similar to using a guide to solve a puzzle. As the conversation goes on, it becomes clear chemical biologists will play a central role in breaking down existing barriers to study things like RNA modifications, PPIs, and more. Near the end, we discuss the relationship between academic science and industry especially as early-career scientists join and start their own companies. Bryan's lab has done incredibly creative research and I would recommend everyone to read his papers. Ultimately, Bryan's work is making an impact in therapeutics with a purview on energy/climate and serves as a template for other inventors looking to build interdisciplinary teams.

Greg is currently a postdoctoral fellow at UCSF studying macrophage immunometabolism in cancer and infection. What makes him truly unique is his ability to do groundbreaking research while training to qualify for the Olympic Trials marathon. Although Greg didn’t make it to the Trials due to injury, he did get a Nature Metabolism publication for his work connecting mitohormesis to immunity, with implications in cancer, longevity, and beyond. Greg grew up in a small Nebraska town where his mother, a kindergarten teacher, cultivated an interest in science. Summer biochemistry research at the University of Colorado spurred Greg to go to UC Berkeley for graduate school at the interface of chemistry and biology. While earning his PhD in the Mark Schlissel & Robert Tjian Labs, Greg would dominate Friday morning MCB basketball games at RSF. He's an even better runner, and was also his high school's starting quarterback. His graduate research focused on the transcriptional regulation of  B cell development. After his PhD, Greg delayed his postdoctoral studies to become lab manager in Kaoru Saijo’s lab when she joined the UC Berkeley faculty to have more time to focus on running. In our conversation, we discuss a mysterious observation that led him down a new thread of immunology. After helping set up the lab, Greg decided to transition from lab manager to postdoc, as a screen of endogenous estrogen metabolites had unveiled an entirely new area of macrophage biology. Greg's work, published in Nature Metabolism, established hydroxyestrogens as anti-inflammatory estrogens that work via inducing mitohormesis and rewiring macrophage metabolism. For background, hormesis essentially means a little bit of stress is positive while too much is negative. Mitohormesis is just this phenomenon in mitochondria, and is best-known as a lifespan-promoting stress response in yeast and worms. It's an incredible story of discovery and connecting mitohormesis to the restraint of inflammation, especially in the backdrop of training for the Olympic Trials. Greg is extending this work at UCSF by exploring an immunometabolic strategy to help macrophages fight cancer and pathogens. Along with this, he has co-founded a startup to translate his mitohormesis work and design novel anti-inflammatory therapeutics. The best is yet to come as Greg defines new areas of immunology and pursues their applications.

Keisuke Goda has set the standard for sorting cells based on images. As a professor at the University of Tokyo leading a lab of 53 scientists, Keisuke is developing new tools to explore biology at different scales. During grad school at MIT, he studied gravitational waves in the LIGO group where a collaboration at Caltech led to an opportunity to move into biology at UCLA. By bringing LIGO technology, and a physicist's perspective, to cytometry, Keisuke has been able to pioneer the field of sorting cells by an image. In our conversation, we touch upon his career starting in Japan, 15 years in the US, and finally back to Japan. Starting around 2007, Keisuke was working on high-speed imaging then moved toward cell sorting after trying to figure out how to integrate imaging into FACS. In short, previous sorting methods don't capture high-resolution spatial features and by adding morphological and other spatial features, we might be able to discover and isolate cells we haven't studied before. Molecular biology has a long history of physicists transforming biology - Keisuke is no different. He made a lot of friends and led a large team of collaborators bringing various fields together from engineering and machine learning to biology. The challenges and opportunities were in managing an interdisciplinary project that had technical roadblocks along the way but more importantly cultural ones. Just as much as Keisuke is world-class at translating different scales of biology, he is equally skilled at translating the languages of various scientific fields. In 2018, the Goda Lab published research in Cell that established image-activated cell sorting (IACS). This led to the creation of an ecosystem of various groups around the world improving the technology and new applications. And the current state-of-the-art is at sorting thousands of cells per second with IACS with around a 10 ms lag time. Throughput can be increased with tradeoffs in purity and imaging success rate with current projects around developing better neural networks and imaging processing methods to increase the number of cells per second that can be sorted. Keisuke pursues data-driven research and treats biology as an open field when new techniques and ideas can be mixed-and-matched together. His lab is pursuing new projects related to merging cytometry, microscopy, and sequencing. Essentially, 3 different languages with each at a different scale of time and space. A key lesson from Keisuke's career is making friends to bring different expertises in order to solve important problems. With his view of a long-term opportunity to integrate biological data at different scales, Keisuke and his group have been really good at making friends.

Nina is one of the up-and-coming superstars in biotech. Having just earned her PhD in bioengineering at Stanford in the Sunwoo Lab this summer, she has the scientific horsepower and storytelling ability to make a large impact on drug development, business, and the lives of patients. At the age of 8, Nina was diagnosed with an ovarian teratoma. In some ways, that shifted Nina towards a career in science and a mission to cure cancer. But while she was always interested in science, growing up in the suburbs of New York City enabled her to become world-class oboe and bassoon player as well. But in her words, she chose the research path rather than joining a conservatory because "science is a way to help other people [while] music is fun." At Williams College, she studied mathematics and biology and became interested in the progress going on in the cell therapy space, particularly CAR-T. This is around 2014 when companies like Juno Therapeutics, Novartis, and Kite Pharma were engineering patient T-cells to hone in on lymphomas. With a growing interest in bioengineering and the early clinical successes in CAR-T, Nina was compelled to join the LEAP program at Boston University to study bioengineering with a goal of engineering immune cells to target cancer. After applying to a number of graduate programs, Nina ended up at Stanford. She rotated through 5 labs: Irv Weissman, Garry Nolan, Crystal Mackall, Edgar Engleman, and John Sunwoo. These experiences ended up having a major impact on her research where she was able to have the flexibility to pursue the projects she was excited by in cell therapies and use various tools/methods she picked up from the other 4 groups in her graduate research. Early-on in grad school, her cancer relapsed and she was treated again. One thing about Nina is her grit. She was able to still move forward with her work despite all of this. Truly amazing. Her first project ended up not working out perfectly but taught Nina how to choose problems to work on. Michael Fischbach helps teach a class at Stanford, BioE 395, on this topic that Nina and I would recommend checking out. Her next project focused on studying subtypes of NK cells in tumors and generating functional data to figure out if they are a unique class of immune cells or an artifact of single-cell sequencing. After figuring out how to run an experiment with 1 month of prep, around 4 days of work, and then 3 weeks of data acquisition and analysis (shout-out to Gail for the help), Nina was able to discover a novel functionality of a new NK cell type that resembles ieILC1s, which have potent anti-tumor activity. Once she was able to show the experiment worked, Nina did the harrowing experiment another five times to verify her discovery and did an incredible job to combine various tools to map out new NK cell biology. At the end of the conversation, we discuss her new job as Head of Research at ImmuneBridge and the opportunities she sees in cell therapies. Nina is truly a role model to scientists and cancer survivors everywhere.