Generative Models for Graph-Based Protein Design

John Ingraham

Last update: Dec 15, 2022

Related tags

Deep Learning neurips19-graph-protein-design

Overview

Graph-Based Protein Design

This repo contains code for Generative Models for Graph-Based Protein Design by John Ingraham, Vikas Garg, Regina Barzilay and Tommi Jaakkola, NeurIPS 2019.

Our approach 'designs' protein sequences for target 3D structures via a graph-conditioned, autoregressive language model:

Overview

struct2seq/ contains model code
experiments/ contains scripts for training and evaluating the model
data/ contains scripts for building and processing datasets in the paper

Requirements

Python >= 3.0
PyTorch >= 1.0
Numpy

Citation

@inproceedings{ingraham2019generative,
author = {Ingraham, John and Garg, Vikas K and Barzilay, Regina and Jaakkola, Tommi},
title = {Generative Models for Graph-Based Protein Design},
booktitle = {Advances in Neural Information Processing Systems}
year = {2019}
}

Comments

#Bug

Dear author,

In your github , the neurips19-graph-protein-design has a build_chain_dataset.py file under the data directory, and a file named mmtf_utils.py in the same directory, but you code

from util_mmtf import *

in the build_chain_dataset.py, it seems not in keeping with the file name mmtf_utils.py。

Maybe a bug?Or i misunderstand something?

Many thanks, Jingru Gao

opened by SharplessSword 0
Why invariance instead of equivariance?

Dear author,

In your paper, the local reference frame was used for invariant edge information. Could you explain why you used invariance instead of equivariance?

Many thanks, Lixin Yang

opened by colormeblue1013 0

Bugs in StructureLoader

There might be a minor bug in data.StructureLoader. Some samples will not be included in the loader, so the dataset size is not exactly the same as that reported in the paper.

for ix in sorted_ix:
    size = self.lengths[ix]
    if size * (len(batch) + 1) <= self.batch_size:
        batch.append(ix)
        batch_max = size
    else:
        clusters.append(batch)
        batch, batch_max = [], 0

opened by veghen 0

no 'GraphAttention' function implementation

class SequenceModel(nn.Module): def init(self, num_letters, hidden_dim, num_layers=3, vocab=20, top_k=30, num_positional_embeddings=16): """ Graph labeling network """ super(SequenceModel, self).init()

    # Hyperparameters
    self.top_k = top_k
    self.hidden_dim = hidden_dim
    self.positional_embeddings = PositionalEmbeddings(num_positional_embeddings)

    # Embedding layers
    self.W_e = nn.Linear(num_positional_embeddings, hidden_dim, bias=True)
    self.W_s = nn.Embedding(vocab, hidden_dim)

    # Decoder
    self.decoder_layers = nn.ModuleList([
        **GraphAttention(hidden_dim, hidden_dim*3)**
        for _ in range(2 * num_layers)
    ])
    self.W_out = nn.Linear(hidden_dim, num_letters, bias=True)

    # Initialization
    for p in self.parameters():
        if p.dim() > 1:
            nn.init.xavier_uniform_(p)

opened by jiyanbio 1

no data/cath

runfile('/Users/hmh2017/Downloads/neurips19-graph-protein-design-master/experiments/train_s2s.py', wdir='/Users/hmh2017/Downloads/neurips19-graph-protein-design-master/experiments') Number of parameters: 1526164 Traceback (most recent call last):

File "/Users/hmh2017/Downloads/neurips19-graph-protein-design-master/experiments/train_s2s.py", line 21, in dataset = data.StructureDataset(args.file_data, truncate=None, max_length=500)

File "../struct2seq/data.py", line 16, in init with open(jsonl_file) as f:

FileNotFoundError: [Errno 2] No such file or directory: '../data/cath/chain_set.jsonl'

opened by minghuihuang 1

Owner

John Ingraham

Postdoc

GitHub

Implementation of the GVP-Transformer, which was used in the paper "Learning inverse folding from millions of predicted structures" for de novo protein design alongside Alphafold2

GVP Transformer (wip) Implementation of the GVP-Transformer, which was used in the paper Learning inverse folding from millions of predicted structure

19 May 6, 2022

Codes and models for the paper "Learning Unknown from Correlations: Graph Neural Network for Inter-novel-protein Interaction Prediction".

GNN_PPI Codes and models for the paper "Learning Unknown from Correlations: Graph Neural Network for Inter-novel-protein Interaction Prediction". Lear

2 Dec 14, 2022

Graph-based community clustering approach to extract protein domains from a predicted aligned error matrix

Using a predicted aligned error matrix corresponding to an AlphaFold2 model , returns a series of lists of residue indices, where each list corresponds to a set of residues clustering together into a pseudo-rigid domain.

24 Nov 23, 2022

Predicting lncRNA–protein interactions based on graph autoencoders and collaborative training

Predicting lncRNA–protein interactions based on graph autoencoders and collaborative training Code for our paper "Predicting lncRNA–protein interactio

1 Nov 29, 2022

DeepCAD: A Deep Generative Network for Computer-Aided Design Models

DeepCAD This repository provides source code for our paper: DeepCAD: A Deep Generative Network for Computer-Aided Design Models Rundi Wu, Chang Xiao,

85 Dec 31, 2022

Unofficial TensorFlow implementation of Protein Interface Prediction using Graph Convolutional Networks.

[TensorFlow] Protein Interface Prediction using Graph Convolutional Networks Unofficial TensorFlow implementation of Protein Interface Prediction usin

9 Oct 25, 2022

This is the repository for the AAAI 21 paper [Contrastive and Generative Graph Convolutional Networks for Graph-based Semi-Supervised Learning].

CG3 This is the repository for the AAAI 21 paper [Contrastive and Generative Graph Convolutional Networks for Graph-based Semi-Supervised Learning]. R

12 Oct 28, 2022

Uni-Fold: Training your own deep protein-folding models

Uni-Fold: Training your own deep protein-folding models. This package provides an implementation of a trainable, Transformer-based deep protein foldin

187 Jan 4, 2023

RITA is a family of autoregressive protein models, developed by LightOn in collaboration with the OATML group at Oxford and the Debora Marks Lab at Harvard.

RITA: a Study on Scaling Up Generative Protein Sequence Models RITA is a family of autoregressive protein models, developed by a collaboration of Ligh

69 Dec 22, 2022

Single Red Blood Cell Hydrodynamic Traps Via the Generative Design

Rbc-traps-generative-design - The generative design for single red clood cell hydrodynamic traps using GEFEST framework

4 Jun 16, 2022

A Protein-RNA Interface Predictor Based on Semantics of Sequences

PRIP PRIP：A Protein-RNA Interface Predictor Based on Semantics of Sequences installation gensim==3.8.3 matplotlib==3.1.3 xgboost==1.3.3 prettytable==2

0 Mar 25, 2022

This project is based on RIFE and aims to make RIFE more practical for users by adding various features and design new models

This project is based on RIFE and aims to make RIFE more practical for users by adding various features and design new models. Because improving the PSNR index is not compatible with subjective effects, we hope this part of work and our academic research are independent of each other.

190 Jan 8, 2023

Generative Models for Graph-Based Protein Design

Related tags

Overview

Graph-Based Protein Design

Overview

Requirements

Citation

Comments

#Bug

Why invariance instead of equivariance?

Bugs in StructureLoader

no 'GraphAttention' function implementation

no data/cath

Owner

John Ingraham

Implementation of the GVP-Transformer, which was used in the paper "Learning inverse folding from millions of predicted structures" for de novo protein design alongside Alphafold2

Codes and models for the paper "Learning Unknown from Correlations: Graph Neural Network for Inter-novel-protein Interaction Prediction".

Graph-based community clustering approach to extract protein domains from a predicted aligned error matrix

Predicting lncRNA–protein interactions based on graph autoencoders and collaborative training

DeepCAD: A Deep Generative Network for Computer-Aided Design Models

Unofficial TensorFlow implementation of Protein Interface Prediction using Graph Convolutional Networks.

This is the repository for the AAAI 21 paper [Contrastive and Generative Graph Convolutional Networks for Graph-based Semi-Supervised Learning].

Uni-Fold: Training your own deep protein-folding models

RITA is a family of autoregressive protein models, developed by LightOn in collaboration with the OATML group at Oxford and the Debora Marks Lab at Harvard.

Single Red Blood Cell Hydrodynamic Traps Via the Generative Design

A Protein-RNA Interface Predictor Based on Semantics of Sequences

This project is based on RIFE and aims to make RIFE more practical for users by adding various features and design new models

RefineGNN - Iterative refinement graph neural network for antibody sequence-structure co-design (RefineGNN)

A denoising diffusion probabilistic model (DDPM) tailored for conditional generation of protein distograms

7th place solution of Human Protein Atlas - Single Cell Classification on Kaggle

Implementation and replication of ProGen, Language Modeling for Protein Generation, in Jax

Replication attempt for the Protein Folding Model

A geometric deep learning pipeline for predicting protein interface contacts.

A package to predict protein inter-residue geometries from sequence data