May 2026—The fast-changing world of sequencing was the topic that a panel of pathologists and industry representatives dug into in a Feb. 27 online roundtable, led by CAP TODAY publisher Bob McGonnagle. Keeping up with the technology is one of the challenges today. Another is “finding a way, as an organized specialty, to keep up with the interoperability of the new data formats we’re acquiring,” said Ulysses G. J. Balis, MD, of Michigan Medicine. The conversation follows.
CAP TODAY’s guide to NGS systems starts here.
Jeremy Segal, in last year’s roundtable you talked about a top-of-mind challenge as you thought about the year ahead: how best to use the different types of panels to optimize patient care and the need for faster turnaround time. Can you update us on your progress on turnaround time and your latest thinking on the panels?
Jeremy Segal, MD, PhD, vice chair for clinical genomics and professor of pathology, University of Chicago: We’ve focused long term on trying to shave time off all steps of the process because patients need the information from our tests as quickly as possible. But at the forefront of my mind this year are all the new things coming down the pike. In particular, it’s been a fascinating year for artificial intelligence. There are many opportunities for AI across pathology in general but also specifically molecular pathology. I’ve spent the past few months working on a system to help support our variant interpretation workflows using artificial intelligence. Things that didn’t seem possible a year ago are now not only achievable but critically time sensitive. In addition to AI, new sequencing technologies, methylation analysis, and multiomics approaches are becoming ready for prime time in the clinical space. How do we evaluate all of these new platforms and make judicious choices about bringing them into the lab? It’s daunting to think about how to approach all of this new technology and how we can balance the competing pressures of turnaround time on the one hand with the demand for increasingly complex and comprehensive biomarker results on the other.
José Luis Costa, what’s your take on AI applications in variant interpretation in molecular pathology? Some studies suggest it is not yet part of standard practice.
José Luis Costa, PhD, global director of scientific affairs, clinical next-generation sequencing and oncology, Thermo Fisher Scientific: It’s well aligned with developments we are trying to bring through our partnerships with OpenAI and Nvidia, for example. AI is a strong part of Thermo Fisher Scientific’s activities throughout our instrumentation and technology portfolios, not just NGS.
With increased sequencing capability, there may be a space for AI to support variant interpretation, helping with filtering, algorithm optimization, and workflows to improve turnaround time. For AI to blindly provide final decisions and interpretations, the methodologies and validations would need to mature. But for optimizing processes and filtering the massive amounts of data that are produced, and integrating data, AI will play a stronger role. It already helps the workflows in which it has been implemented.
Pierre Del Moral, do you agree that AI right now is showing its best mettle in workflow as opposed to providing interpretation or definitive answers?
Pierre Del Moral, PhD, MBA, global senior director of health care customer marketing, Illumina: To me it’s with data and data interpretation. Illumina launched BioInsight, a business that focuses on data analysis to meet the demand for deeper biologic insights. It goes beyond the data generated in a diagnostic setting; it’s working with pharma to make sense of the data and improve the diagnostic yield. How do we leverage the massive amount of data generated in clinical trials, for example? How do we tie it together? How do we integrate multiomics so it’s more than one-dimensional? How does one interpret and make the right decision for therapies based on multimodal testing results from NGS, digital pathology, and digital PCR?