Protein Pathogenicity Analysis Tool

This project offers an accessible script to analyze and visualize AlphaMissense-predicted pathogenicity scores. It integrates with ClinVar and AlphaFold to promote a more comprehensive interpretation of the pathogenicity predictions.

Features

• Fetches and processes AlphaMissense predictions for a specified protein (based on UniProt ID)
• Automatically fetches AlphaFold PDB files or utilizes user-provided experimental PDB files to generate a modified PDB file with AlphaMissense pathogenicity scores
  • Replaces the temperature factor (B-factor) with pathogenicity values, enabling visualization in molecular visualization tools
• Creates a heatmap with the AlphaMissense predicted pathogenicity scores for all amino acid residue substitutions
• Extracts reporter missense variants from ClinVar
  • Compares the pathogenicity classification of all missense ClinVar variants to the AlphaMissense classifications (exported as {GENE_ID}_clinvar_AM.csv)
• Produces a line graph visualizing:
  • Average AlphaMissense predicted pathogenicity at each amino acid residue
  • AlphaFold per-residue model confidence score (pLDDT)
  • Secondary structure annotations for alpha helices and beta sheets
• Produces another line graph visualizing:
  • Average AlphaMissense predicted pathogenicity at each amino acid residue
  • Missense variants from ClinVar along with their associated classification (simplified for plot simplicity)

Requirements

• Python 3.11 or higher

You can install the required packages using:

pip install -r requirements.txt

Alternatively, you can create a conda environment with the required packages using:

conda create --name amissense python requests pandas seaborn matplotlib plotly numpy biopython conda-forge::python-kaleido salilab::dssp
conda activate amissense

Installation

With the setup.py file, you can now install the package using the following:

pip install .

This will install the amissense package and make the amissense command-line interface (CLI) available.

Usage

To run the main pipeline, use the following command:

amissense pipeline -u UNIPROT_ID -g GENE_ID [-o OUTPUT_DIR] [-e EXPERIMENTAL_PDB]

Arguments:

• UNIPROT_ID: The UniProt ID of the protein you want to analyze. This is a required positional argument.
• GENE_ID: The gene ID associated with the protein. This is a required positional argument.
• OUTPUT_DIR: The directory to store output files (default is out/).
• EXPERIMENTAL_PDB: The experimental PDB file for the protein (optional). If not provided, the script will use the AlphaFold predicted structure.

Example Command

amissense pipeline -u P12345 -g BRCA1 -o output_dir -e experimental.pdb

Utility Commands

You can also use utility commands for additional operations:

• Download AlphaMissense predictions:

amissense utils download-predictions -t /path/to/tmp_dir

• Download a PDB file:

amissense utils download-pdb -p 6LID -o /path/to/output_dir

• Query UniProt for a gene's UniProt ID:

amissense utils uniprot-query -n GENE_NAME -i ORGANISM_ID

JSON Generation

To generate JSON files from AlphaMissense TSV data:

amissense generate-json /path/to/AlphaMissense_aa_substitutions.tsv.gz /path/to/output_dir

Output

The script generates several output files in the specified output directory {GENE_ID}_{UNIPROT_ID}_{YYYY-MM-DD}:

Figures

• {UNIPROT_ID}_heatmap.png: Heatmap of the AlphaMissense pathogenicity scores for each amino acid substitution
• {UNIPROT_ID}_line_graph.png: Line graph showing the AlphaMissense mean pathogenicity, AlphaFold per-residue model confidence score (pLDDT), and secondary structure annotations
• {GENE_ID}_avgAM_clinvar.png: Line graph showing the AlphaMissense mean pathogenicity and extracted ClinVar missense variants with their classification
• {GENE_ID}_sankey_diagram.png: Sankey diagram depicting the flow quantity between the AlphaMissense variant pathogenicity classification and the ClinVar variant classifications

PDB

• {UNIPROT_ID}_alphafold_{YYYY-MM-DD}.pdb: Unaltered PDB file from AlphaFold
• {UNIPROT_ID}_pathogenicity.pdb: PDB file with pathogenicity scores replacing the B-factor

Tables

• {UNIPROT_ID}_AM_pathogenicity_predictions_{YYYY-MM-DD}.csv: CSV file containing the AlphaMissense predictions
• {GENE_ID}_clinvar_AM_{YYYY-MM-DD}.csv: CSV file with ClinVar and AlphaMissense data combined