Scaling Proteomics with Milad Dagher

What It Covers

Milad Dagher, cofounder and CEO of Nomic Bio, explains how the Analyzer platform scales multiplexed ELISA-based proteomics to 200 proteins per sample using pre-assembled antibody pairs on color-coded beads, enabling drug discovery teams to run high-throughput protein measurements at costs low enough for routine daily use across large sample cohorts.

Key Questions Answered

• •Multiplexing architecture: Nomic solves the cross-reactivity problem that limits traditional multiplex immunoassays by pre-assembling each antibody pair onto individual color-coded beads before the assay begins. This spatial localization eliminates antibody mixing entirely, unlike proximity extension assays which allow cross-reactivity and then discriminate against it. The result is consistent signal quality whether running 10-plex or 200-plex panels, enabling direct data comparison across experiments.
• •Cost-driven discovery scale: Most biomarker discovery studies using OLink or SomaLogic measure thousands of proteins but only across a few hundred samples, creating overfitting risk. Nomic's lower cost structure enables population-scale studies at 10,000+ samples. Researchers can run broad 200-plex discovery panels first, then build smaller custom panels for the 30–100 proteins of interest, compounding cost efficiency for validation-stage work.
• •Drug discovery target validation: Nomic ran a single experiment profiling 10,000 wells — PBMCs from six donors, stimulated four ways, perturbed with 60 immunomodulatory cytokines — measuring 200 proteins per well. This generated thousands of protein interaction data points in one experiment, a dataset that would have taken years with standard ELISA. Teams using this approach have identified unexpected off-target signals during lead optimization that single-biomarker readouts would have missed entirely.
• •Service-first go-to-market: Launching as a service rather than a kit allowed Nomic to iterate on antibody panel quality, learn directly from scientist workflows, and validate product-market fit within months instead of years. OLink followed a similar path and now derives roughly half its revenue from kits. Scientists who initially requested kits often prefer staying on the service after experiencing the workflow simplicity, which also provides Nomic ongoing data on assay performance across diverse applications.
• •Proteomics vs. transcriptomics ground truth: RNA sequencing data frequently does not correlate with actual protein expression, yet much published biology relies on transcriptomic proxies. Proteomics platforms like the Analyzer provide direct protein quantification, making them a more reliable basis for target validation and drug development decisions. Researchers should treat RNA-seq leads as hypotheses requiring protein-level confirmation before committing program resources to a specific target or pathway.
• •Founder learning cadence: Dagher credits Y Combinator and Creative Destruction Lab primarily for the founder and mentor networks rather than curriculum. He identifies self-awareness about role-specific skill gaps as the most critical CEO capability, noting the job effectively resets every six months as the company scales. Structured self-evaluation routines and executive coaching become necessary once peer learning from cofounders and accelerator networks no longer covers the new demands of each growth stage.