Mendelspod Podcast

For decades, drug developers have struggled with so-called "undruggable" proteins—those regions of the proteome that evade traditional small molecules and antibodies. But Aikium, led by Eswar Iyer, could be changing the game. In today’s show, Iyer, a prolific scientist with over 100 patents and a background in George Church’s lab, shares how his company is tackling one of the toughest challenges in therapeutics.At the heart of Aikium’s approach is Yotta-ML2, an AI-powered wet lab platform that searches the vast combinatorial space of proteins to create precise, bespoke binders for disease targets that have long been out of reach. “The combinatorial space for how many proteins can bind a region is very large,” Iyer explains. “Experiments are limited to 10⁹ to 10¹² possibilities, but the theoretical space exceeds 10¹⁵. We’re using AI to intelligently navigate this massive search problem and find what actually works in the body.”Iyer is clear-eyed about the biggest challenges ahead: “AI is just hype if it can’t deliver better therapies, faster,” he says. “The biggest barrier isn’t just designing binders—it’s making sure they work in human biology, avoiding immunogenicity, and accelerating the slow feedback cycle of drug development.”Looking to the future, Iyer envisions a major shift in the industry: “We’re living in an age of exponential technologies. Just as AI has transformed other fields, it’s going to compress the timeline and risk of therapeutic development. In five or six years, we’ll see a wave of AI-designed molecules entering the clinic.” This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.mendelspod.com/subscribe

Today Theral talks with Laura Hercher, Director of Research for Human Genetics at Sarah Lawrence College, about the growing political assault on diversity, equity, and inclusion (DEI) initiatives in scientific research. With the NIH facing sweeping budget cuts and DEI programs cancelled, Hercher makes a compelling case for why diversity isn’t just a social or moral issue—but a scientific imperative."What you're cutting is cancer funding. What you're cutting is Alzheimer's research. What you're cutting is funding for sick kids," Hercher warns, pushing back against the rhetoric that frames these cuts as mere "overhead reductions."She breaks down how diversity strengthens both scientific databases and the research workforce, emphasizing that genetic studies are incomplete without diverse representation. She also addresses the political motivations behind these attacks, highlighting the broader pattern of anti-intellectualism and anti-science sentiment creeping into public policy.Hercher closes with a powerful statement about her own institution’s commitment to DEI:"We are not changing a damn thing. We think it's right, and we're sticking to it."Join us today and make your own voice heard if you’re concerned about the future of scientific research and equity in medicine.Editor’s Note: Today’s show is dedicated to our late Executive Producer, Ayanna Monteverdi. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.mendelspod.com/subscribe

This is a free preview of a paid episode. To hear more, visit www.mendelspod.comLung cancer is the most deadly of all the cancers worldwide. Very few of those who should be— are getting screened. There is enormous medical need here, and early cancer detection through advanced liquid biopsy could be a tremendous part of the answer.Today, as part of our series on the future of genetic testing, we welcome Susan Tousi, CEO of Delfi Diagnostics, a company pioneering next-generation liquid biopsy technology for early cancer detection. Tousi came to the post a year ago formerly serving as Illumina’s Chief Commercial Officer. Delfi has made significant strides in developing affordable, high-sensitivity blood-based tests focusing on lung cancer.Tousi discusses the company’s groundbreaking approach, which leverages whole-genome sequencing and machine learning to detect cancer at its earliest stages."We believe that for early detection to make a difference and make cancer less deadly, you have to be catching it at stage one or stage two," she emphasizes.By using a low-pass sequencing method, Delfi is able to keep costs low while maximizing sensitivity, making widespread screening more accessible. The conversation highlights the dire need for better lung cancer screening methods, given that fewer than 6% of eligible individuals undergo annual screening with low-dose CT scans."Lung cancer alone takes more lives annually than colorectal, breast, and cervical cancer combined," Tousi notes.Delfi aims to change this with its simple blood test, already in commercial use and being adopted by major healthcare systems such as OSF HealthCare and Allegheny Health Network.Beyond lung cancer, Tousi also shares insights into Delfi’s broader vision, including its work on liver and ovarian cancer detection, as well as its potential in tumor monitoring."Every day that goes by without better screening means lives lost that we could have saved,” she states.

This is a free preview of a paid episode. To hear more, visit www.mendelspod.comIn our first conversation with Deep Genomics, Theral sits down with Brendan Frey, the company's founder and Chief Innovation Officer, to explore the power of AI in RNA biology and drug discovery. Frey, a pioneer in deep learning who trained in the lab of AI luminary Geoffrey Hinton, shares how his personal experience in genetics back in 2002 led him to merge machine learning with genomic medicine. The company has a head start.In the past year, Deep Genomics has built Big RNA, the first foundational model for RNA biology, trained on over a trillion genomic signals. "RNA is the ideal point of intervention," says Frey. "It’s the software of biology—it takes context into account, shaping how DNA functions across different cells and conditions."

In this Mendelspod episode, Theral explores the state of genome informatics with Ben Busby, principal scientist at DNAnexus.Busby reflects on his journey into the field, which was influenced by key projects and experiences during his time at the NCBI and his genomics hackathons. He outlines the current genome informatics landscape, emphasizing DNAnexus' role in fostering collaboration. Busby highlights transformative movements to refine disease subtyping based on haplotype data, underscoring their potential to reshape diagnostics and research.As for the use of the new generative AI in bioinformatics, Busby says, "It is not super good at making data into information. And it’s not particularly good at making information into knowledge, but it does understand something about making knowledge into impact.”Looking ahead, he envisions a future where genome informatics drives precision medicine and fosters global collaborations leveraging more diverse genomic data. He ends with a call for longitudinally resolved databases. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.mendelspod.com/subscribe

This is a free preview of a paid episode. To hear more, visit www.mendelspod.comAs 2025 begins, we are launching a new series The New Genetic Testing. To do that we’re welcoming today Dr. Ezra Cohen, Chief Medical Officer at Tempus to offer a behind-the-scenes look at how the company is transforming healthcare through cutting-edge AI. With its recent successful IPO, Tempus has firmly established itself as a leader in genomic testing and multimodal data integration and empowering providers and life sciences companies to make better decisions for patients."Tempus is really a tech company applying sophisticated AI to the healthcare space," explains Dr. Cohen. With tools like TempusOne, an AI-powered assistant, oncologists and other clinicians can access real-time insights, from interpreting genetic tests to exploring treatment options and clinical trials.”Dr. Cohen highlights Tempus’ unique approach to data collection, amassing 300 petabytes of multimodal data—50 times larger than the largest public oncology database. This vast resource, paired with advanced AI, allows Tempus to provide unparalleled context for each patient."We’re not just saying this cancer has this gene mutated. We’re layering genetic mutations, co-mutations, expression levels, radiographs, digital pathology, and treatment outcomes to give providers a comprehensive picture," he says.As Tempus continues to expand its offerings and impact, Dr. Cohen envisions a future where AI reshapes every aspect of healthcare, from patient interactions to drug development."The future is disruptive, and it's going to change incredibly fast. Fasten your seatbelts—we’re in an amazing time."

This is a free preview of a paid episode. To hear more, visit www.mendelspod.comThis week we welcome the innovative minds behind PredixBio, a new and groundbreaking company at the forefront of spatial biology. CEO Dusty Majumdar and co-founder Chakra Chennubhotla join us to discuss their mission to revolutionize cancer therapeutics through spatially intelligent biology. Founded in 2018 and supported by NIH grants, PredixBio focuses on unraveling the tumor microenvironment's heterogeneity using cutting-edge analytics and explainable AI. Their SpaceIQ platform integrates images and transcriptomics to predict patient responses to immunotherapy with an accuracy of over 90%.Chakra studied under AI pioneer Geoffrey Hinton, often referred to as the "godfather" of AI. He explains the transformative potential of spatial analytics."Heterogeneity is not random; there is an order to it. Understanding this order allows us to uncover mechanisms hidden in the data."Dusty highlights the company's mission, emphasizing the importance of explainability in AI for clinical trials and drug discovery."We don’t just provide predictions; we explain why patients respond—or don’t—to therapies. This helps redefine new targets and improves outcomes."Join in to find out how PredxBio's new approach is setting new standards and shaping the future of spatial biology and precision medicine.

This year in a series co-produced with GenomeWeb we had an exclusive look into Illumina’s work in genomic AI with Kyle Farh, the VP and Distinguished Scientist leading the Illumina Artificial Intelligence Lab. "We lack a way of training genomic AI as readily as you would for natural language processing,” says Farh. "There are vast amounts of data, but there are no labels, no supervision. One of the most powerful tools that we’ve found is using clues from natural selection.”Farh says the company began the AI lab seven years ago when deep learning took off. Two of the team’s AI algorithms, PrimateAI-3D and SpliceAI, published in Cell and Nature Genetics, were among the first examples of deep learning algorithms in genomics to become widely used by clinical labs and researchers in interpreting variants of unknown significance.What will be the long-term impact of AI on genomics?Chapters:0:00 First-of-kind studies to utilize AI to decode variation in humans3:00 Training genomic AI on natural selection9:10 What sets Illumina’s algorithms apart?11:40 Insights of PrimateAI-3D and SpliceAI21:40 How will AI impact genomics long term? This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.mendelspod.com/subscribe

This is a free preview of a paid episode. To hear more, visit www.mendelspod.comIn this episode of Mendelspod, Theral sits down with Dr. Mohamad Takwa, co-founder and CEO of Epigenica, a Swedish startup revolutionizing epigenetic research. Epigenica is positioning itself at the forefront of this new and rapidly growing field with a mission to empower researchers through large-scale epigenetic profiling.

Coming from the field of digital pathology, Joachim Schmid has a unique perspective on the evolution of multi-omics.  Schmid was recently appointed as Vice President of Multiomics Data Solutions at Illumina.In this episode, Theral dives into the burgeoning field of multi-omics—integrating data from various omic layers such as genomics, proteomics, and transcriptomics—and the massive data challenges that come with it. Schmid discusses how sequencing costs are plummeting, leading to an explosion in multi-omics applications. Yet, with all the data being generated, biologists often lack the computational background to make sense of it.  Schmid highlights Illumina's acquisition of Partek, a company that has been developing multi-omics data analysis tools for over 30 years, and he emphasizes that empowering researchers with accessible analysis tools is crucial to advancing the field.AI is poised to play a key role in moving multi-omics forward. Schmid discusses both the opportunities and limitations of AI, noting that while AI can streamline data processing, there are still areas for improvement, particularly in developing more intuitive and user-friendly tools for biologists. He shares some exciting applications of multi-omics in personalized medicine and cancer research. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.mendelspod.com/subscribe