Data

Tools

Indexes

Applications

Experiments

Open Source Science

protein-detective

Python package to detect proteins in EM density maps.

https://github.com/haddocking/protein-detective

Science Score: 67.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ✓
    CITATION.cff file
    Found CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ✓
    DOI references
    Found 3 DOI reference(s) in README
  • ✓
    Academic publication links
    Links to: zenodo.org
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (11.9%) to scientific vocabulary
Last synced: 6 months ago · JSON representation ·

Repository

Python package to detect proteins in EM density maps.

Basic Info
Statistics
  • Stars: 0
  • Watchers: 4
  • Forks: 0
  • Open Issues: 10
  • Releases: 4
Created 11 months ago · Last pushed 6 months ago
Metadata Files
Readme Contributing License Code of conduct Citation

README.md

protein-detective

Documentation CI Research Software Directory Badge PyPI DOI

Python package to detect proteins in EM density maps.

It uses

  • Uniprot Sparql endpoint to search for proteins and their measured or predicted 3D structures.
  • powerfit to fit protein structure in a Electron Microscopy (EM) density map.

An example workflow:

mermaid graph LR; search{Search UniprotKB} --> |uniprot_accessions|fetchpdbe{Retrieve PDBe} search{Search UniprotKB} --> |uniprot_accessions|fetchad{Retrieve AlphaFold} fetchpdbe -->|mmcif_files| residuefilter{Filter on nr residues + write chain A} fetchad -->|pdb_files| densityfilter{Filter out low confidence} residuefilter -->|pdb_files| powerfit densityfilter -->|pdb_files| powerfit powerfit -->|*/solutions.out| solutions{Best scoring solutions} solutions -->|dataframe| fitmodels{Fit models}

Install

shell pip install protein-detective

Or to use the latest development version: pip install git+https://github.com/haddocking/protein-detective.git

Usage

The main entry point is the protein-detective command line tool which has multiple subcommands to perform actions.

To use programmaticly, see the notebooks and API documentation.

Search Uniprot for structures

shell protein-detective search \ --taxon-id 9606 \ --reviewed \ --subcellular-location-uniprot nucleus \ --subcellular-location-go GO:0005634 \ --molecular-function-go GO:0003677 \ --limit 100 \ ./mysession (GO:0005634 is "Nucleus" and GO:0003677 is "DNA binding")

In ./mysession directory, you will find session.db file, which is a DuckDB database with search results.

To retrieve a bunch of structures

shell protein-detective retrieve ./mysession

In ./mysession directory, you will find mmCIF files from PDBe and PDB files and AlphaFold DB.

To filter AlphaFold structures on confidence

Filter AlphaFoldDB structures based on confidence. Keeps entries with requested number of residues which have a confidence score above the threshold. Also writes pdb files with only those residues.

shell protein-detective confidence-filter \ --confidence-threshold 50 \ --min-residues 100 \ --max-residues 1000 \ ./mysession

To prune PDBe files

Make PDBe files smaller by only keeping first chain of found uniprot entry and renaming to chain A.

shell protein-detective prune-pdbs \ --min-residues 100 \ --max-residues 1000 \ ./mysession

Powerfit

Rotate and translate the prepared structures to fit and score them into the EM density map using powerfit.

shell protein-detective powerfit run ../powerfit-tutorial/ribosome-KsgA.map 13 docs/session1

This will use dask-distributed to run powerfit for each structure in parallel on multiple CPU cores or GPUs.

Run powerfits on Slurm You can use [dask-jobqueue](https://jobqueue.dask.org/en/latest/) to run the powerfits on a Slurm deployment on multiple machines on a shared filesystem. In one terminal start the Dask cluster with ```shell pip install dask-jobqueue python3 ``` ```python from dask_jobqueue import SLURMCluster cluster = SLURMCluster(cores=8, processes=4, memory="16GB", queue="normal") print(cluster.scheduler_address) # Prints something like: 'tcp://192.168.1.1:34059' # Keep this Python process running until powerfits are done ``` In second terminal, run the powerfits on Dask cluster with ```shell protein-detective powerfit run ../powerfit-tutorial/ribosome-KsgA.map 13 docs/session1 --scheduler-address tcp://192.168.1.1:34059 ```
How to run efficiently Powerfit is quickest on GPU, but can also run on CPU. To run powerfits on a GPU you can use the `--gpu `. The value of `workers_per_gpu` should be high enough so the GPU is fully utilized. You can start with 1 (the default) and monitor the GPU usage with `nvtop` if you see that the GPU is not 100% loaded, you can increase the number until there are no more valleys in the GPU usage graph. If you have multiple GPUs, then `--gpu 2` will run powerfits on all GPUs and run 2 powerfits concurrently on each GPU. If you do not use `--gpu` flag, then powerfit will run on CPU. By default each powerfit will use 1 CPU core and run multiple powerfits in parallel according to the number of physical CPU cores available on the machine (so excluding hyperthreaded cores). You can set the `--nproc ` so each powerfit will use that many CPU cores. This is useful if you have more CPU cores available then there are structures to fit. If the number of structure to fit is greater than available CPU cores then using the default (1 core per powerfit) is recommended.
Alternativly run powerfit yourself You can use the `protein-detective powerfit commands` to print the commands. The commands can then be run in whatever way you prefer, like sequentially, with [GNU parallel](https://www.gnu.org/software/parallel/), or as a [Slurm array job](https://slurm.schedmd.com/job_array.html). For example to run with parallel and 4 slots: ```shell protein-detective powerfit commands ../powerfit-tutorial/ribosome-KsgA.map 13 docs/session1 > commands.txt parallel --jobs 4 < commands.txt ```

To print top 10 solutions to the terminal, you can use:

shell protein-detective powerfit report docs/session1

Outputs something like:

powerfit_run_id,structure,rank,cc,fishz,relz,translation,rotation,pdb_id,pdb_file,uniprot_acc 10,A8MT69_pdb4e45.ent_B2A,1,0.432,0.463,10.091,227.18:242.53:211.83,0.0:1.0:1.0:0.0:0.0:1.0:1.0:0.0:0.0,4E45,docs/session1/single_chain/A8MT69_pdb4e45.ent_B2A.pdb,A8MT69 10,A8MT69_pdb4ne5.ent_B2A,1,0.423,0.452,10.053,227.18:242.53:214.9,0.0:-0.0:-0.0:-0.604:0.797:0.0:0.797:0.604:0.0,4NE5,docs/session1/single_chain/A8MT69_pdb4ne5.ent_B2A.pdb,A8MT69 ...

To generate model PDB files rotated/translated to PowerFit solutions, you can use:

shell protein-detective powerfit fit-models docs/session1

Contributing

For development information and contribution guidelines, please see CONTRIBUTING.md.

Owner

  • Name: HADDOCK
  • Login: haddocking
  • Kind: organization
  • Location: Utrecht, The Netherlands

Computational Structural Biology Group @ Utrecht University

Citation (CITATION.cff) 

# YAML 1.2
---
cff-version: 1.2.0
title: protein detective
message: >-
  If you use this software, please cite it using the
  metadata from this file.
type: software
authors:
  - affiliation: "Netherlands eScience Center"
    family-names: Verhoeven
    given-names: Stefan
    orcid: "https://orcid.org/0000-0002-5821-2060"
  - given-names: Anna
    family-names: Engel
    affiliation: '@UtrechtUniversity'
    orcid: "https://orcid.org/0009-0002-1806-7951"
  - given-names: Alexandre
    family-names: Bonvin
    affiliation: '@UtrechtUniversity'    
    orcid: "https://orcid.org/0000-0001-7369-1322"
repository-code: https://github.com/haddocking/protein-detective
identifiers:
  - description: Latest version of software
    type: doi
    value: 10.5281/zenodo.15632658

GitHub Events

Total
  • Create event: 17
  • Release event: 1
  • Issues event: 23
  • Delete event: 15
  • Member event: 2
  • Issue comment event: 20
  • Push event: 74
  • Pull request review comment event: 3
  • Pull request review event: 3
  • Pull request event: 29
Last Year
  • Create event: 17
  • Release event: 1
  • Issues event: 23
  • Delete event: 15
  • Member event: 2
  • Issue comment event: 20
  • Push event: 74
  • Pull request review comment event: 3
  • Pull request review event: 3
  • Pull request event: 29

Issues and Pull Requests

Last synced: 6 months ago

All Time
  • Total issues: 19
  • Total pull requests: 23
  • Average time to close issues: 8 days
  • Average time to close pull requests: 6 days
  • Total issue authors: 2
  • Total pull request authors: 1
  • Average comments per issue: 1.16
  • Average comments per pull request: 0.09
  • Merged pull requests: 15
  • Bot issues: 0
  • Bot pull requests: 0
Past Year
  • Issues: 19
  • Pull requests: 23
  • Average time to close issues: 8 days
  • Average time to close pull requests: 6 days
  • Issue authors: 2
  • Pull request authors: 1
  • Average comments per issue: 1.16
  • Average comments per pull request: 0.09
  • Merged pull requests: 15
  • Bot issues: 0
  • Bot pull requests: 0
View more stats
Top Authors
Issue Authors
  • sverhoeven (18)
  • rvhonorato (1)
Pull Request Authors
  • sverhoeven (21)
Top Labels
Issue Labels
Pull Request Labels

Packages

  • Total packages: 1
  • Total downloads:
    • pypi 20 last-month
  • Total dependent packages: 0
  • Total dependent repositories: 0
  • Total versions: 4
  • Total maintainers: 1
pypi.org: protein-detective

Deduce the protein from a EM density

  • Versions: 4
  • Dependent Packages: 0
  • Dependent Repositories: 0
  • Downloads: 20 Last month
Rankings
Dependent packages count: 9.0%
Average: 29.9%
Dependent repos count: 50.8%
Maintainers (1)
bonvinlab
Last synced: 6 months ago