Benchmarking RNA Inverse Folding

A robust, open-source framework for RNA inverse design in 2D and 3D. Unified environments, curated datasets, and state-of-the-art algorithms.

Start Designing Explore Data

Secondary Design

Design sequences from secondary structures including complex pseudoknots. Benchmarked against diverse datasets including Eterna100v2 and Rfam.

Tertiary Design

Reconstruct RNA sequences from full 3D PDB structures. Includes specialized processing for converting 3D coordinates into algorithmic tensors.

Web Server

Immediate access to 12+ machine learning algorithms. Design sequences, optimize G-C content, and minimize free energy without local installation.

About The Project

Through RNAInvBench, we aim to promote reproducible, comparable, and extensible benchmarking for the global RNA design research community.

Task Diversity

Covers both secondary and tertiary design. Secondary design consists of four core tasks (including pseudoknots and constraints), whereas tertiary design focuses on reverse-engineering 3D structures.

Algorithm Baselines

We provide easy access to a large collection of RNA secondary and tertiary inverse folding algorithms through an integrated web server.

Benchmark Datasets

Includes large pseudoknot-free and pseudoknot-inclusive sets (Rfam, Eterna100v2), plus curated tertiary structure benchmarks from RNAsolo and CASP15.

Who is this for?

  • Researchers developing novel inverse RNA design methods
  • Computational biologists seeking standardized benchmarks
  • Bioinformaticians working on RNA structure prediction
  • Developers integrating or comparing RNA design algorithms

Where to start?

UoE Logo JIC Logo