ProteinFold

Institut Curie - Nextflow ProteinFold analysis pipeline

Introduction

The pipeline is built using Nextflow, a workflow tool to run tasks across multiple compute infrastructures in a very portable manner. It comes with singularity/apptainer containers making installation easier and results highly reproducible.

Pipeline summary

This pipeline allows:

• the prediction of protein 3D structure using various tools:
  • AFMassive.
  • AlphaFold,
  • AlphaFold3,
  • ColabFold,
• molecular docking of protein/ligand using:
  • DiffDock
  • DynamicBind
  • AlphaFill
• the navigation of the nanobody mutational space
  • nanoBERT

Quick help

nextflow run main.nf --help

```bash N E X T F L O W ~ version 22.10.6

Launching main.nf [curious_perlman] DSL2 - revision: 986ad6e9f0

  _   _   _   _   _   _   _     _   _   _   _  
 / \ / \ / \ / \ / \ / \ / \   / \ / \ / \ / \ 
( P | r | o | t | e | i | n ) ( F | o | l | d )
 \_/ \_/ \_/ \_/ \_/ \_/ \_/   \_/ \_/ \_/ \_/ 

Usage:

The typical command for running the pipeline is as follows:

nextflow run main.nf -params-file params.json -profile STRING 

MANDATORY ARGUMENTS, NEXTFLOW: -profile STRING [test, singularity, cluster] Configuration profile to use. Can use multiple (comma separated).

OTHER OPTIONS, NEXTFLOW: -params-file PATH Set the parameters of the pipeline using a JSON file configuration file (i.e. 'params.json'). All parameters defined as JSON type must be this way. For example, the JSON can contain: "alphaFoldOptions": "--max_template=2024-01-01 --multimer". WARNING: passing the option '--alphaFoldOptions' in command line will throw an error when the option contains '-' or '--' characters which are not appreciated by nextflow.

OTHER OPTIONS: --afMassiveDatabase PATH Path to the database required by AFMassive. --afMassiveHelp Display all the options available to run AFMassive. Use this option in combination with -profile singularity. afMassiveOptions JSON Specific options for AFMassive. As AFMassive is an AlphaFold-like tool, standard AlphaFold options are passed using the --alphaFoldOptions option. --alphaFillHelp Display all the options available to run AlphaFill. Use this option in combination with -profile singularity. --alphaFold3Database PATH Path to the database required by AlphaFold3. --alphaFold3Help Display all the options available to run AlphaFold3. Use this option in combination with -profile singularity. alphaFold3Options JSON Prediction model options passed to AlphaFold3. --alphaFoldDatabase PATH Path to the database required by AlphaFold. --alphaFoldHelp Display all the options available to run AlphaFold. Use this option in combination with -profile singularity. alphaFoldOptions JSON Prediction model options passed to AlphaFold or AFMassive. --colabFoldDatabase PATH Path to the database required by ColabFold. --colabFoldHelp Display all the options available to run ColabFold. Use this option in combination with -profile singularity. colabFoldOptions JSON Prediction model options passed to ColabFold. --diffDockArgsYamlFile YAML Path to the YAML file with the DiffDock options. --diffDockDatabase PATH Path to the database required by DiffDock. --dynamicBindDatabase PATH Path to the database required by DynamicBind. --dynamicBindHelp Display all the options available to run DynamicBind. Use this option in combination with -profile singularity. dynamicBindOptions JSON Prediction model options passed to DynamicBind. --fastaPath PATH Path to the input directory which contains the fasta files. --fromMsas PATH Path to existing multiple sequence alignments (msas) to use for the 3D protein strcuture prediction. Typically the path could be the results of the pipeline launcded with the --onlyMsas option. --launchAfMassive Launch AFMassive --launchAlphaFill Launch AlphaFill. --launchAlphaFold Launch AlphaFold. --launchAlphaFold3 Launch AlphaFold3. --launchColabFold Launch ColabFold. --launchDiffDock Launch DiffDock. --launchDynamicBind Launch DynamicBind. --multimerVersions INT AlphaFold multimer model versions (v1, v2, v3) which will be evaluated by AFMassive. This parameter is taken into account when --launchAfMassive is true. The list of the versions to be evaluated must be provided with a comma separated string, e.g. 'v1,v2', Default is 'v1,v2,v3'. --numberOfModels INT Number of models that will be evaluated by AFMassive. This parameter is taken into account when --launchAfMassive is true. --onlyMsas When true, the pipeline will only generate the multiple sequence alignments (msas). --outDir PATH The output directory where the results will be saved --predictionsPerModel INT Number of predictions per model which will be evaluated by AFMassive. This parameter is taken into account when --launchAfMassive is true. --proteinLigandFile PATH Path to the input file for molecular docking. The file must be in CSV format, without space. One column named 'protein' contains the path the the 'pdb' file and one column named 'ligand' must contain the path to the 'sdf' file. --useGpu Run the prediction model on GPU. AlphaFold and AFMassive can run either on CPU or GPU. ColabFold and DynamicBind require GPU only.

REFERENCES: --genomeAnnotationPath PATH Path to genome/proteome annotations folder used to predict the protein 3D structure.

======================================================= Available Profiles -profile test Run the test dataset -profile singularity Use the Singularity images for each process. Use --singularityPath to define the insallation path

-profile cluster Run the workflow on the cluster, instead of locally

```

Quick run

The pipeline can be run on any infrastructure. The use of GPU is preferred (and even required for some tools) to speed-up computation.

Run the pipeline on a test dataset

See the conf/test.config to set your test dataset.

nextflow run main.nf -profile test,singularity

Run the pipeline with custom option

Create a json file with your custom parameters, for example:

json { "alphaFoldOptions": "--max_template_date=2024-01-01 --random_seed=654321" }

Launch nextflow with the -params-file options: nextflow run main.nf --fastaPath="test/data" -params-file params.json --outDir MY_OUTPUT_DIR -profile singularity

Run the pipeline on a computing cluster

For example, to launch the pipeline on a computing cluster with SLURM:

bash echo "#! /bin/bash" > launcher.sh echo "set -oue pipefail" >> launcher.sh echo nextflow run main.nf --fastaPath="test/data" -params-file params.json --outDir MY_OUTPUT_DIR -profile singularity,cluster >> launcher.sh sbatch launcher.sh

Full Documentation

1. Installation
2. Running the pipeline
3. Output of the pipelines

Credits

This pipeline has been written by the bioinformatics platform of the Institut Curie (P. Hupé).

This work was granted access to the HPC resources of IDRIS under the allocation 2024-AD010315541 made by GENCI.

Citation

Contacts

For any question, bug or suggestion, please use the issues system or contact the bioinformatics core facility.