By Diagnostics World Staff
November 11, 2025 | Last week, veterans of Illumina and Grail launched Hepta, a biotechnology company using transformer-based AI to read the cell-free DNA (cfDNA) epigenome and detect organ-specific signals of chronic disease. Hepta emerged from stealth last Wednesday with $6.7 million in seed funding led by Felicis Ventures and Illumina Ventures, with participation from SeaX Ventures, Alumni Ventures, and AME Cloud Ventures.
The company was founded by Hamed Amini, Ph.D. formerly with Illumina and Grail, now Hepta’s CEO and Soheil Damangir, Ph.D. formerly with Grail and Google AI, now Hepta’s CTO.
The company claims to have the first liquid biopsy-native AI platform that detects liver disease with tissue-level biological insights. In May, Abdelmalek et al published clinical data in the Journal of Hepatology showing that its AI-powered epigenetics model can identify patients with metabolic dysfunction-associated steatohepatitis (MASH, formerly known as NASH) with significant fibrosis with a diagnostic AUC of 0.86, cutting false positives threefold compared to standard blood tests (DOI: 10.1016/S0168-8278(25)01466-7)
Yesterday, the company released details about its MASH Atlas dataset, developed in collaboration with Duke University and other leading hepatology centers. The atlas integrates single-cell and bulk data across hundreds of liver samples, including gene expression, chromatin accessibility, and DNA methylation, paired with cfDNA methylation from matched blood plasma. The atlas shows high concordance between cfDNA methylation and liver tissue methylation and gene expression, and reveals how coordinated methylation and transcriptional programs drive inflammation and fibrosis and demonstrates that these same molecular signals are detectable in blood using Hepta’s liquid biopsy-native AI platform.
“Across independent cohorts, we see strong, consistent diagnostic performance from Hepta’s cfDNA methylation platform,” said Jörn Schattenberg, M.D., Director of the Metabolic Liver Research Center at Saarland University, who presented the MASH atlas at the Precision Liver Symposium. “What’s striking is that the same readout also delivers deep biological insights. It’s not just a yes-or-no test; these signals reveal how repair and fibrosis programs evolve, which opens the door to earlier intervention and therapy guidance.”
Hepta believes these strengths further confirm the ability of this technology to act as a true liquid biopsy of the liver, providing a new lens to detect and study the disease as well as discover novel therapeutic targets. Hepta’s CEO Amini posited that the platform also holds potential, “as a single source for guiding therapy and monitoring response.”
Hepta is clinically demonstrating for the first time that epigenetic patterns in circulating cell-free DNA (cfDNA) mirror liver-tissue biology in metabolic dysfunction-associated steatohepatitis (MASH).
MASH, a progressive form of metabolic dysfunction‑associated steatotic liver disease (MASLD, formerly known as NAFLD or non-alcoholic fatty liver disease), affects more than 20 million Americans, yet less than 1% are diagnosed today. Current testing methods, unchanged for two decades, cannot scale to meet this need. Invasive biopsies and specialized imaging like FibroScan exist in fewer than 10% of primary care settings. Blood tests like FIB-4 produce false positives in over half of patients in low-prevalence populations. As the first MASH therapies reach the market with at least seven more in late-stage trials, physicians need tools that work where patients actually are. Hepta's blood-based test promises to deliver tissue-level insight from a simple blood draw, bringing specialist-grade precision to routine care.
Hepta has developed a first-of-its-kind liquid-biopsy-native transformer model optimized for cfDNA analysis. Unlike conventional oncology assays that focus on high-signal ctDNA fragments or limited gene panels, this architecture analyzes the epigenetic patterns across all the molecules in the blood sample in a single pass, capturing contextual relationships among up to a billion molecular interactions. The company's patented platform leverages the attention mechanism, the core innovation behind large language models like GPT, scaling it to one billion unique molecules per sample. This allows the model to track how every observed molecule interacts with every other molecule, uncovering subtle, disease-specific biological patterns.
"Traditional liquid biopsy methods work for cancer by typically looking for rare but distinct mutations, chromosomal copy number variations, or rather unique epigenetic changes," said CTO Damangir in a press release. "Chronic diseases such as MASH create no such signals, but rather subtle methylation shifts across millions of genomic sites. The technical challenge is detecting complex diseases like MASH, where signals are far fainter than in cancer. We built a transformer that processes all billion molecular interactions simultaneously. That's how we find patterns that are invisible to methods that analyze the genome in far smaller segments."
"Hepta is extending the power of liquid biopsy beyond oncology to reveal organ-level biology in chronic disease," added Amini, in the same press release. "Our team, which developed next generation sequencing technologies at Illumina and built and scaled the first generation of liquid biopsy platforms at Grail, has now demonstrated that a simple blood draw can replicate tissue-level biology. This establishes a foundation for non-invasive diagnostics that can guide precision therapies in chronic diseases.”