Use Proto Online
Commercially Available Proto No-Code Web Server
Proto: A High-Level Programming Language for Generative Biology Across DNA, RNA and Protein Design
Proto is a high-level programming language for generative biology that lets you compose design campaigns across DNA, RNA, proteins and ligands from a small set of building blocks. It was developed by Aditi T. Merchant, Daniel Guo, Ben Viggiano, Lucas Brennan-Almaraz, Evelyn Hur, Tina Mai, Peter Yin, Samuel H. King, Euan A. Ashley and Brian L. Hie at Stanford University and the Arc Institute. The authors describe it as an open resource, and the Proto tools library is open source under the MIT License.
Biological design is still laborious and heuristic, leaning on natural parts, intuition, screening and repeated build-test-learn cycles. Existing generative tools are specialized: protein design mostly targets individual molecules, generative genomics is constrained to natural genetic architectures, and language models are bounded by existing knowledge and can hallucinate. No unified abstraction composed these tools across modalities and scales.
Attribute | Detail |
|---|---|
Primitives | Sequences, constraints, generators, optimizers |
Formal basis | Energy-based model, target proportional to p(x) exp(-f(x)/T) |
Default optimizer | Metropolis-Hastings MCMC with simulated annealing |
Tool library | Wraps over 120 tools, each in an isolated environment |
Interfaces | Python API, web GUI (proto.evodesign.org), agentic MCP interface |
How Proto Composes Generative Biology Design Campaigns
A Proto program is built from four primitives, so the same pattern applies whether you design a protein, an intron or a promoter.
Sequences: typed DNA, RNA, protein or ligand strings.
Constraints: scalar scoring functions, which can wrap a neural network such as AlphaFold. Multiple constraints are summed, equivalent to a product of experts.
Generators: proposal distributions such as language models, diffusion models or random samplers.
Optimizers: loops that accept, reject or rank candidates, including MCMC with simulated annealing, rejection sampling, gradient-based methods, a cycling optimizer and beam search.
Tool ecosystem
Proto wraps public models including Evo 2, ESM3, ESMFold, Boltz-2, AlphaFold 3, AlphaGenome, Borzoi, Enformer, ProteinMPNN, LigandMPNN and RFdiffusion3, with automatic weight provisioning and GPU scheduling.
Experimental and Computational Results Designed with Proto
Campaign | Result |
|---|---|
ProtoIntrons (65 tested) | 32% of successfully assayed designs showed significant differential splicing in the designed direction |
ProtoPromoters | 45 of 86 gave over 10-fold eGFP induction above the no-promoter control |
ProtoRepressors | 46% significantly reduced eGFP across 12 of 14 promoter targets |
CRISPR-Cas loci | 40 passing designs from 48,000 sampled loci (in silico) |
Human complexes | 249 complexes diversified; median RMSD 1.9 A, TM-score 0.80 (in silico) |
The authors also recapitulated earlier campaigns in silico, including symmetric homo-oligomers from trimers to octamers, a 20-kb chromatin sequence encoding PROTO (AUROC 0.98) and Germinal PD-L1 binders.
What is Tamarind Bio?
Tamarind Bio is a no-code bioinformatics platform built to give life scientists and researchers access to powerful computational tools. Many cutting-edge machine learning models are hard to deploy and use. Tamarind provides an intuitive, web-based environment that removes the complexity of high-performance computing, software dependencies and command-line interfaces.
The platform is designed for biologists, chemists and other researchers who may not have a background in programming or cloud infrastructure but want to run models on their own data. Key features include:
A user-friendly graphical interface for setting up and launching experiments
A robust API for integration into existing research pipelines
An automated system for managing and scaling computational resources
Tamarind treats information and data security as a top priority, as detailed in its Trust Center and Terms of Service.
Accelerating Discovery with Proto on Tamarind Bio
Regulatory element design: generate introns, promoters and repressors optimized for a chosen cell context.
Protein and complex design: diversify human protein complexes or design binders with structure-based constraints.
Multi-objective campaigns: combine several scoring functions across modalities in one design loop.
Agent-driven design: let a coding agent drive Proto through its MCP interface for large-scale campaigns.
How to Use Proto on Tamarind Bio
Open the tool: log in to tamarind.bio and open Proto.
Define the design goal: specify the sequence type (DNA, RNA, protein or ligand) and what the design should achieve.
Choose constraints: if exposed, select scoring functions such as structure or expression predictors.
Pick generator and optimizer: if exposed, choose the proposal model and optimizer; simulated annealing is the default described by the authors.
Run the campaign: submit the job and wait for candidates to be generated and ranked.
Download outputs: save the ranked candidate sequences and scores.
Validate: test top designs experimentally, since computed scores are plausibility filters.
Things to Keep in Mind
Experimental validation was low-N, with tens of designs per system.
Structure-prediction metrics are plausibility filters, not guarantees of function, and some results are computational only.
No single constraint separated functional from non-functional introns and repressors, so composite scoring was needed.
One objective (HepG2 mis-spliced, K562 correct) produced no significantly differential designs.
Optimizers are approximate, and more efficient methods are needed.
Source: Merchant AT, Guo D, Viggiano B, Brennan-Almaraz L, Hur E, Mai T, Yin P, King SH, Ashley EA, Hie BL. A high-level programming language for generative biology with Proto. bioRxiv, 2026, DOI 10.64898/2026.06.22.733870, https://www.biorxiv.org/content/10.64898/2026.06.22.733870v1.full.pdf. Web app: https://proto.evodesign.org/.