# fold2seq

**Repository Path**: mirrors_ibm/fold2seq

## Basic Information

- **Project Name**: fold2seq
- **Description**: Code for Fold2Seq paper from ICML 2021
- **Primary Language**: Unknown
- **License**: Apache-2.0
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-06-09
- **Last Updated**: 2025-12-14

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# [ICML2021] Fold2Seq: A Joint Sequence(1D)-Fold(3D) Embedding-based Generative Model for Protein Design

![Fold2Seq Architecture](/fold2seq1.png)

## Environment file: 
* [environment.yml](environment.yml)

## Data and Feature Generation:
* Go to [data/](data/) and check the README there. 

## How to train the model:
* go to [src/](src/) and run:

`python train.py --data_path $path_to_the_data_dictionary --lr $learning_rate --model_save $path_to_the_saved_model`

## How to generate sequences:
* go to [src/](src/) and run:

`python inference.py --trained_model $path_to_the_trained_model --output $path_to_the_output_file --data_path $path_to_the_data_dictionary`


## Fold2Seq generated structures against natural structures:
![Fold2Seq structures](/fold2seq3.png)