Trained with 12M protein structures predicted by AlphaFold2, the ESM-IF1 model consists of invariant geometric input processing layers followed by a sequence-to-sequence transformer, and achieves 51% native sequence recovery on structurally held-out backbones with 72% recovery for buried residues.

Trained with 12M protein structures predicted by AlphaFold2, the ESM-IF1 model consists of invariant geometric input processing layers followed by a sequence-to-sequence transformer, and achieves 51% native sequence recovery on structurally held-out backbones with 72% recovery for buried residues.

This repository contains code and pre-trained weights for Transformer protein language models from Facebook AI Research, including our state-of-the-art ESM-1b and MSA Transformer, as well as ESM-1v for predicting variant effects and ESM-IF1 for inverse folding.

This repository focuses on designing T-cell receptors (TCRs) using the ESM-IF1 deep learning method. The ESM-IF1 inverse folding method is built for predicting protein sequences from their backbone atom coordinates.

This repository contains code and pre-trained weights for Transformer protein language models from Facebook AI Research, including our state-of-the-art ESM-2 and ESMFold, as well as MSA Transformer, ESM-1v for predicting variant effects and ESM-IF1 for inverse folding.

The ESM-IF1 inverse folding model is built for predicting protein sequences from their backbone atom coordinates. We provide scripts here 1) to sample sequence designs for a given structure and 2) to score sequences for a given structure.

Inverse folding is the task of predicting a sequence of amino acids that will produce a given protein backbone structure. We evaluated two different methods for this step, ProteinMPNN and ESM-IF1; we find ProteinMPNN to be significantly more performant.

AntiFold is based on the ESM-IF1 model and is fine-tuned on solved and predicted antibody structures from SAbDab and OAS. Paper: arXiv pre-print; Webserver: OPIG webserver; Colab: Model: model.pt; License: BSD 3-Clause

cannot import name 'esm_if1_gvp4_t16_142M_UR50' Hello, When I started to use ESM2, some problems appeared. Here is my code: model, alphabet = esm.pretrained.esm2_t33_650M_UR50D() And it returned: AttributeError: module 'esm.pretrained' h...

This repository holds inference and training code for ProteinDPO (Protein Direct Preference Optimization), a preference optimized structure-conditioned protein language model based on ESM-IF1. We describe ProteinDPO in the paper “Aligning Protein Generative Models with Experimental Fitness via Direct Preference Optimization” .