Code and Data for: A Zero-Shot Generative Framework for Stable Genetic Circuits

Read full text Summarise with VIRAT
Authors: Abhinav Bhushan
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This repository contains the custom computational pipelines, datasets, and analysis scripts required to reproduce the findings presented in the manuscript "Zero-Shot Foundation Models Predict Evolutionary Stability to Enable In Silico Optimization of Synthetic Promoters." The codebase utilizes the Nucleotide Transformer (v2-500m-multi-species foundation model) to perform strictly zero-shot inference on a combinatorially engineered E. coli MPRA library. By evaluating evolutionary naturalness rather than predicting raw transcriptional output, the pipeline identifies and optimizes promoters for long-term genetic stability and reduced metabolic burden. The repository includes scripts for: Zero-Shot Scoring: Calculating evolutionary fitness scores via cross-entropy loss without assay-specific training data. Statistical Benchmarking: Executing computational baselines, including partial correlation (controlling for GC% bias), 5-mer Ridge Regression, and canonical $\sigma^{70}$ Position Weight Matrix (PWM) scoring. Phenotypic Segregation: Generating 2D Kernel Density Estimate (KDE) topographical maps to isolate optimal "High-Fitness/High-Expression" biomanufacturing candidates. Sensitivity Mapping: Performing in silico saturation mutagenesis to generate heatmaps highlighting model sensitivity to canonical -10 and -35 RNA polymerase binding motifs. Computational Upgrades: Executing AI-driven in silico sequence optimization to computationally alleviate evolutionary friction and upgrade candidate sequences prior to physical synthesis. Requirements: Python 3.9+, PyTorch, HuggingFace Transformers (

Fields of Study & Subjects

Microbial Metabolic Engineering and Bioproduction Gene Regulatory Network Analysis RNA and protein synthesis mechanisms

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
25 March 2026
Language
EN