The development of novel therapies for serious unmet medical needs demands a platform that is simultaneously fast, versatile, and cost-effective. For Decoy Therapeutics, a preclinical-stage biopharmaceutical company, the mission is even more focused: addressing global health threats of existing and novel viruses by applying an engineering mindset to antiviral drug discovery. Founded during the early days of the COVID-19 pandemic, Decoy took inspiration from antibody-drug conjugates to design peptide conjugates to block viruses from infecting human cells. The idea was to essentially turn the Swiss watch complexity of viral fusion machinery into a solvable engineering problem.

To build designable multi-antivirals (D-MAVs™), a new therapeutic class designed to target shared viral mechanisms across multiple viruses, Decoy needed to unify isolated components. Their infrastructure was fragmented: genomic data lived on Compute Engine, protein data resided in local SQL databases, and various workflows were scattered across disconnected systems. The primary pain point was the need to consolidate these pipelines into a single, production-grade platform capable of handling high-performance computing (HPC) for their two major programs.

When selecting a cloud partner, Google Cloud was the ideal choice due to its deep-rooted leadership in life sciences and its foundational role in the very tools Decoy uses daily. "The most important tool for this work is AlphaFold2, which came from Google," explains Peter Marschel, Co-founder and Chief Business Officer. 

For a team leveraging the solution to the once-insolvable "protein folding problem," being able to run these tools on their native infrastructure provided a strategic advantage that made Google Cloud the obvious environment for their research.

Implementing a unified computational platform required shifting from bespoke Python scripts on local machines to an autoscalable, containerized architecture. Working with Google Cloud, Decoy developed a technical infrastructure heavily rooted in Google Kubernetes Engine (GKE), featuring GPU-enabled node pools specifically designed for heavy-duty biological simulations. This system is managed through robust CI/CD pipelines using Cloud Build, Docker, and Artifact Registry, ensuring that the environment is as repeatable as the science itself.

The platform focuses on two core HPC modules that are critical to the drug design process:

1. Molecular Dynamics (MD) Simulation Pipeline: This module utilizes GPU acceleration and optimized engines like GROMACS. To optimize costs, the team uses "Precision Switching"—dynamically choosing between single and double precision—and incorporates machine-learning force fields to approximate interactions, significantly accelerating simulation speed.
2. Tiered Protein Folding Pipeline: Decoy implements structure prediction (forward) and design (reverse) workflows. By exposing a tiered set of algorithms, researchers can trade accuracy for speed depending on the needs of the calculation.

Decoy is increasing their computational bandwidth by 25 to 100-fold using Google Cloud, helping in part to accelerate discovery and optimization timelines that would typically take five to seven years down to a single year. The speed of the computational design, paired with their proprietary "single-shot" synthesis technology, allows them to move from design to physical molecule in hours or days—a process that previously took months.

It demonstrates how a lean team can use Google Cloud AI, HPC, and GKE solutions to build a high-value, scalable business without the need to hire massive teams to manage data centers or security infrastructure.

This efficiency allows Decoy to focus entirely on their mission: creating broad-acting antivirals that are as accessible in low and middle-income countries as they are in the US and Europe.

Looking ahead, Decoy is working with a leading peptide manufacturer to scale-up their proprietary peptide-conjugate synthesis technology to global scale on typical commercially available peptide synthesis equipment. The goal is to be able to take D-MAV candidates designed on the Google Cloud to clinical and commercial scale manufacturing with unprecedented speed, and to easily share that manufacturing globally through a standardized ‘tech transfer’ and licensing process. Put another way, with Google Cloud and others, Decoy Therapeutics is not just building a drug pipeline; they are engineering a faster, more equitable future for global healthcare.