About Us

The genomics revolution was made possible by one thing: a readable molecule. Once biology had a tool that could sequence DNA directly and at scale, the data flowed, and a rich seam of biological innovation followed.

Proteins are a different story. The proteome contains more than one million distinct proteoforms — compared to 20,000 genes in the human genome. It is the most information-rich layer of biology, and until now, the least readable. Not because the data isn't there, but because the tools to extract it haven't existed.

Existing proteomics platforms sample biology — they take limited, targeted measurements from a small, well-lit corner of the proteome. deSIPHR mines it. It reads proteins directly, without a reference catalog, at a throughput and cost trajectory that makes comprehensive biological data extraction not just possible but systematic.

The pumpkinseed is a small, brightly colored freshwater fish native to North America, known for the iridescent, jewel-like patterns on its scales. Pumpkinseed's nanophotonic chip wafers catch the light in much the same way as a pumpkinseed's scales: vivid, structured, and unexpectedly beautiful. It is also a nod to the idea of something small that contains the potential for something much larger.

Pumpkinseed was co-founded by three scientists whose disciplines don't usually share a lab and whose convergence is what made the platform possible. Prof. Jen Dionne (Stanford Engineering, Chan-Zuckerberg Biohub) whose world-leading research in nanophotonics and single-cell Raman spectroscopy is the scientific backbone of deSIPHR. Dr. Jack Hu (Forbes 30 Under 30, inventor on ten patents) translated that science into a working instrument, one amino acid at a time. Dr. Nhat Vu brings 15 years of machine learning experience from Google, Nest, and Jawbone, and leads the computational infrastructure that turns raw proteomic signal into biological insight.

deSIPHR is Pumpkinseed's proprietary nanophotonic chip platform for measuring protein biology. With over 100 million sensors per chip it is designed to interrogate proteins across multiple dimensions - from molecular identity and modification state to molecular interactions.

For sequencing, deSIPHR reads molecular building blocks one at a time using Raman spectroscopy. Rather than relying on a reference protein, Raman reads molecular vibrations directly. The resulting physical signatures contain rich information about molecular identity, structure, and interactions.

Because deSIPHR chips are fabricated using semiconductor manufacturing, the platform is designed to scale the way silicon scales: more sensors, higher throughput, and lower cost per data point over time.

De novo protein sequencing means determining a protein's amino-acid sequence directly, without requiring its identity or sequence to be known in advance.

Most proteomic workflows identify proteins indirectly—for example, by fragmenting molecules and interpreting the resulting spectra, often with the aid of reference databases. deSIPHR takes a different approach: using Raman spectroscopy to read molecular identity directly from vibrational signatures, one building block at a time.

Raman spectroscopy reads the molecular vibrations of individual molecules, revealing rich information about their identity, structure, and interactions.

deSIPHR combines Raman spectroscopy with nanophotonics and molecular chemistry to amplify and interpret these otherwise faint vibrational signatures. This is physics reading biology in the most literal sense.

deSIPHR is a platform technology with applications wherever richer measurements of protein biology could change what's possible.

In biopharma, we're interested in protein and peptide discovery, protein–small molecule interactions, biologics characterization, and drug-response studies. In biosecurity, direct molecular measurement could help characterize unfamiliar or engineered biological threats. For BioAI, deSIPHR is designed to generate linked experimental data describing what proteins are, what they interact with, and what happens next.

Across these applications, the goal is the same: move from molecular identity to interaction, from interaction to function, and ultimately from function to prediction.

Genentech's Discovery Proteomics group is an active biopharma partner, where our work in immunopeptidomics is already revealing peptide targets that mass spectrometry cannot detect. We hold active contracts with DARPA and BARDA across defense and biosecurity applications, and have a manufacturing and fabrication relationship with IMEC.

We are actively partnering with teams where direct protein measurement would change what's possible in their programs. That includes immunopeptidomics programs in cancer vaccine development, autoimmune research, and biologics immunogenicity; process analytic teams monitoring biologic drug manufacturing; and synthetic biology groups engineering cells as biological factories, where the protein output of engineered cell lines is complex, novel, and poorly characterized by existing methods.

If there is a protein measurement you wish you could make, we'd like to hear about it.

We work with partners on defined proteomics programs. Partners bring samples, biological questions, and domain expertise; our team contributes the platform, experimental design, and the analytical infrastructure to translate proteomic findings into actionable insights. Early partnerships are structured as pilots, with the opportunity to deepen into broader data-sharing and co-development arrangements as the platform scales.

We are headquartered in San Carlos, CA. Our team of physicists, biologists, engineers, chemists, and data scientists is growing. For current opportunities, contact us at info@pumpkinseed.bio.