🐠 at Future Ventures. The colorful wafer contains billions of optical nanosensors for sequencing peptides and proteins. Each amino acid in the protein chain is read sequentially from its unique vibrational signature. Protein sequencing has proven to be a more difficult challenge than DNA sequencing, but Pumpkinseed made it accurate and fast with a huge silicon photonics array performing Raman spectroscopy in parallel. This allows analysis of the proteome with all of its post-translational complexity, a new lens into the inner workings of our biology.

The company is named Pumpkinseed after the most colorful freshwater fish in North America, bearing a likeness to their chips, from the diffraction grating effect of their sensor arrays.

I put my 8” wafer of 100 MHz Intel Pentiums next to it for comparison (below). Gordon Moore gave that to me in 1994, and when I was a student of Andy Grove, he was kind enough to sign it.

• Some more background: otl.stanford.edu/sites/g/files/sbiybj16766/files/media/fi…
• Inquiries: pumpkinseed.bio

3 responses to “Pumpkinseed Bio wafer of nano photonics sensors and founding team”

  1. Geeking out on this with my new Nueralink-GrokAI-Human-Intelligence-Enhancement-Module 😉 >>

    Protein sequencing determines the linear order of amino acids in a peptide or protein chain (its primary structure). This is typically done experimentally on the actual protein/peptide molecule, rather than solely from genetic predictions.

    In the context of emerging optical nano-sensor approaches (particularly plasmonic nanopore-enhanced Raman spectroscopy or SERS), it often involves reading amino acids sequentially by their unique vibrational (Raman) signatures as the molecule translocates through or interacts with nanoscale hotspots. Each of the 20 standard amino acids (and many post-translational modifications) produces distinct Raman spectra due to differences in their chemical bonds, side chains, and vibrational modes.

    How Optical Nano-Sensors (Plasmonic Nanopores + Raman) Work for Sequencing

    These systems combine nanopores (tiny apertures, often solid-state or hybrid) with plasmonic nanostructures (e.g., gold nanoparticles or bowl-shaped pores) that create intense electromagnetic "hotspots" for surface-enhanced Raman scattering (SERS).

    Process overview:
    Proteins/peptides are often linearized or unfolded and driven (e.g., electrophoretically) through or near the plasmonic nanopore. As each amino acid passes through the optical hotspot, the plasmonic enhancement massively boosts the weak Raman signal (by factors of 10^6–10^10 or more), enabling single-molecule detection.
    High-speed Raman spectrometers capture vibrational spectra (fingerprints) with microsecond resolution.
    Machine learning/AI classifies the spectra to identify individual amino acids or residues sequentially.

    Demonstrated capabilities include:
    Discriminating all 20 proteinogenic amino acids at the single-molecule level.
    Detecting single amino acid substitutions or post-translational modifications (PTMs) in peptides.
    Label-free operation (no fluorescent tags needed, unlike some other methods).

    This is an active research area (e.g., projects like RamanProSeq) aiming for true single-molecule, label-free protein sequencing, analogous to how nanopore tech revolutionized DNA sequencing but using optical readout for richer chemical information.

    Note that most current nanopore protein work is electrical (measuring ionic current blockades), but optical/plasmonic variants add spectroscopic detail for better amino acid discrimination.

    Why Sequence Proteins Directly Instead of (or in Addition to) Inferring from Genetic Code?

    The genetic code (DNA/RNA → mRNA codons → amino acid sequence during translation) predicts the primary sequence as synthesized on the ribosome. However, the actual protein in a cell often differs:

    Post-translational modifications (PTMs): Phosphorylation, glycosylation, acetylation, ubiquitination, etc., alter amino acids after synthesis. These are critical for function, localization, stability, and signaling but are invisible in the genome/transcriptome.

    Alternative splicing, RNA editing, or proteolytic processing: These create protein isoforms or truncated versions not directly predictable from a single gene sequence.
    Proteoforms and variants: Real-world proteins include mutations, degradation products, or environmental influences. Sequencing the actual molecule captures what’s present and functional.

    Non-template synthesis or unknown genes: Some peptides (e.g., in microbes or novel organisms) lack clear genomic templates.

    Clinical/diagnostic value: Direct sequencing identifies disease-specific proteoforms, biomarkers, or misfolded proteins (e.g., in cancer, neurodegeneration) that genomics alone misses. It also verifies predictions and quantifies expression/PTMs at the protein level.

    Bottom line: Genomics gives the blueprint; direct protein sequencing reveals the executed, modified product. Optical nano-sensors promise high-throughput, single-molecule resolution for this, potentially enabling proteomics at scales and speeds rivaling next-gen DNA sequencing, with advantages in label-free chemical specificity via vibrational signatures. These technologies are still emerging but show strong proof-of-concept for peptides and are progressing toward full proteins.

  2. ps: kind of amazed at the concept of Post-translational modifications (PTMs) (Phosphorylation, glycosylation, acetylation, ubiquitination, etc.), that alter amino acids after synthesis. This is a new one for me…and the idea that the genome determines a protein’s nominal form such that it can be modified in-situ by various factors to suit critical run-time function is very surprising…a whole additional realm of complexity and flexibility in the life.exe engine…wow.

Leave a Reply

Your email address will not be published. Required fields are marked *