Ep158: David Younger and Randolph Lopez on SynBio & AI for Discovery

Episode Summary

Executive Summary: A-AlphaBio co-founders David Younger and Randolph Lopez explain how their Seattle biotech uses yeast-based synthetic biology, AlphaSeq, and machine learning to measure protein-protein interactions at massive scale. The platform supports both partnered programs and internal drug discovery, especially molecular glues and cis-targeting immunocytokines, aiming to make protein drug discovery faster, cheaper, and more predictive.

Main Topics: Founders’ backgrounds and partnership (Priority: 5/5): David grew up in Seattle in a medical family and studied bioengineering; Randolph came from Caracas, moved to the U.S. in 2007, worked at Illumina, then pursued a UW PhD. Their complementary technical and operational skills became the basis for the company. Origin of AlphaSeq at the University of Washington (Priority: 5/5): The platform began as an academic effort combining synthetic biology, protein engineering, and yeast genetics at UW’s Institute for Protein Design and Center for Synthetic Biology, with an early focus on turning sequencing into a readout for protein interactions. How AlphaSeq works (Priority: 5/5): Engineered A and alpha yeast cells are stripped of native mating proteins, programmed to express proteins of interest, and then used to detect binding through cell fusion. Sequencing barcodes provides quantitative readouts of interaction strength at million-scale throughput. AI/ML as an analysis and design engine (Priority: 5/5): Machine learning became essential once the team generated enough data that human analysis was impractical. Models are trained on AlphaSeq datasets to improve prediction, reduce blind alleys, and accelerate iterative protein design. Partnered drug discovery programs (Priority: 4/5): A-Alpha has collaborated with Gilead, Amgen, and Bristol-Myers Squibb. Gilead’s project focuses on broadly neutralizing HIV biologics, while Amgen and BMS partnerships focus on molecular glues and E3 ligase-targeting discovery. Internal pipeline and company strategy (Priority: 5/5): Rather than operate as a CRO, A-Alpha uses partnerships to validate the platform while building proprietary programs in cis-targeting immunocytokines and molecular glues, with a goal of retaining value and creating long-term drug assets. Capital strategy and company-building path (Priority: 4/5): As first-time founders without major VC access, they pursued a stepwise path: de-risk the technology, secure IP, publish/partner selectively, and use collaborations as external validation before scaling financing.

Key Arguments: Large-scale protein-protein interaction data are the bottleneck in protein therapeutics; AlphaSeq solves this by measuring millions of interactions in one assay. Next-generation sequencing enabled a new kind of synthetic biology experiment where sequencing acts as a readout for biological function, not just genomes. Machine learning is most useful when paired with high-quality, diverse experimental data; better data directly improves prediction and speeds design cycles. A yeast system is valid for discovery because it is eukaryotic, benchmarked, and can identify which human proteins are likely to express and behave well. Partnerships were essential for de-risking the platform, creating references, and providing near-term milestones for early investors. A-Alpha is differentiated not just by data volume but by its ability to measure weak interactions and use mutational analysis to inform molecular glue discovery. The company’s internal focus is on areas where its platform offers a unique edge and strong market need: cytokine engineering and rational molecular glue discovery. AI will significantly improve drug discovery and optimization, but it will not yet solve downstream clinical trial attrition or human biology uncertainty.

Data Points: Company founded: 2017 - A-AlphaBio was founded as a spin-out from the University of Washington. Randolph moved to the U.S.: 2007 - He came from Caracas, Venezuela, to attend college and pursue bioengineering. Randolph joined Illumina: 2012 - He worked in engineering after graduating before returning to academia. Randolph started UW PhD: 2013 - He moved to Seattle to pursue a PhD and work with UW protein engineering leaders. David finished PhD: 2017 - He completed his doctoral work and then helped transition the project toward a company. CoMotion postdoc support: 1 year - UW commercialization support gave David time to de-risk the business without publication pressure. Proprietary interaction database: Over 750 million - A-Alpha’s owned database of measured protein-protein interactions has grown to this scale. AlphaSeq scale per assay: Thousands by thousands - They described routine assays measuring about 1,000 proteins against another 1,000 proteins. Modeling outcome: One or two iterations instead of three or four - They said aggregated data can accelerate antibody affinity maturation cycles. Ligase universe: ~600 ligases - They noted the human proteome contains about 600 described E3 ligases. Academic-to-company support: Multiple UW business plan competitions - These helped pressure-test the founders’ working relationship and business plan.

Pivotal Quotes: "we can measure millions of interactions between proteins quantitatively all in a single assay" — Randolph Lopez: Explaining the core AlphaSeq capability and why it matters for protein drug discovery. "the bottleneck is moving those computational designs and testing them experimentally" — Randolph Lopez: Describing why protein design needed a high-throughput experimental platform. "we knew that if we just open sourced and, you know, wrote a paper and let this go out into the public domain, then there would be really nobody who would have the right incentives to put in the investment" — David Younger: Explaining why the team chose commercialization over a purely academic publication path.

Implications: A-AlphaBio shows how deep experimental data plus ML can reshape biologics discovery. For biotech, the message is clear: platforms that generate proprietary ground truth at scale can support both partnered revenue and high-value internal drug programs.