Running AlphaFold2 (from ColabFold) in Azure Machine Learning

Colby T. Ford, Ph.D.

Companion repository for Medium Post: How to predict many protein structures with AlphaFold2 at-scale in Azure Machine Learning.

Resources:

Jupyter Notebook: AzureML_AlphaFold2_HyperDrive.ipynb
Sample Sequence FASTA: sequences.fasta
Dockerfile: Dockerfile

Docker Image Creation

The custom Docker image for running ColabFold in Azure ML HyperDrive is available on Docker Hub at https://hub.docker.com/r/cford38/alphafold2_aml. If you'd like to build the image for yourself, follow the instructions below.

First, build the custom Azure ML Docker image (with the required dependencies for the AlphaFold2 system).

docker build -t alphafold2_aml --build-arg IMAGE_VERSION=$(git rev-parse --short HEAD) .

# docker run --name alphafold2_aml --rm -p 8787:8787 alphafold2_aml
# docker exec -it alphafold2_aml /bin/bash

To push the image to DockerHub, run the following:

docker image tag alphafold2_aml <USERNAME>/alphafold2_aml:latest
docker push <USERNAME>/alphafold2_aml:latest

# docker image tag alphafold2_aml cford38/alphafold2_aml:latest
# docker push cford38/alphafold2_aml:latest

To push the image to an Azure Container Registry, run the following:

az login
az account set --subscription <SUBSCRIPTION_ID>

az acr login --name <CONTAINER_REGISTRY_NAME>

docker tag mmae_aml <CONTAINER_REGISTRY_NAME>.azurecr.io/alphafold2_aml:$(git rev-parse --short HEAD)
docker push <CONTAINER_REGISTRY_NAME>.azurecr.io/alphafold2_aml:$(git rev-parse --short HEAD)

Running AlphaFold2 (from ColabFold) in Azure Machine Learning

Related tags

Overview

Running AlphaFold2 (from ColabFold) in Azure Machine Learning

Colby T. Ford, Ph.D.

Resources:

Docker Image Creation

Owner

Colby T. Ford

Self-Supervised Monocular 3D Face Reconstruction by Occlusion-Aware Multi-view Geometry Consistency[ECCV 2020]

System Combination for Grammatical Error Correction Based on Integer Programming

PyTorch implementation of "Efficient Neural Architecture Search via Parameters Sharing"

Detectron2-FC a fast construction platform of neural network algorithm based on detectron2

Official implementation for paper: A Latent Transformer for Disentangled Face Editing in Images and Videos.

Add-on for importing and auto setup of character creator 3 character exports.

Keras-retinanet - Keras implementation of RetinaNet object detection.

StyleTransfer - Open source style transfer project, based on VGG19

implementation for paper "ShelfNet for fast semantic segmentation"

AI drive app that can help user become beautiful.

Implementation of Barlow Twins paper

NALSM: Neuron-Astrocyte Liquid State Machine

A simple and lightweight genetic algorithm for optimization of any machine learning model

Pre-trained Deep Learning models and demos (high quality and extremely fast)

Generic Foreground Segmentation in Images

codes for Image Inpainting with External-internal Learning and Monochromic Bottleneck

Keras udrl - Keras implementation of Upside Down Reinforcement Learning

A brand new hub for Scene Graph Generation methods based on MMdetection (2021). The pipeline of from detection, scene graph generation to downstream tasks (e.g., image cpationing) is supported. Pytorch version implementation of HetH (ECCV 2020) and TopicSG (ICCV 2021) is included.

Deep Inertial Prediction (DIPr)

Jupyter Dock is a set of Jupyter Notebooks for performing molecular docking protocols interactively, as well as visualizing, converting file formats and analyzing the results.