izzetbiophysicist / VCF_context

Fetches functional annotation, AlphaMissense Prediction and generates structures for variants on a VCF file

VCF_context

This tool is designed to fetch functional annotations and AlphaMissense predictions for missense variants found in a VCF file.

Step 1: Annotate VCF with Annovar

1.1. Use the table_annovar.pl script to annotate your VCF file. For example:

perl table_annovar.pl example/ex2.vcf humandb/ -buildver hg38 -out myanno -remove -protocol refGene,cytoBand,exac03,avsnp147,dbnsfp30a -operation g,r,f,f,f -arg '-hgvs',,,, -nastring . -vcfinput -polish

The resulting text file generated in this step will be used in the subsequent steps. Note that, in the following scripts, only missense variants will be considered.

Annovar can be downloaded from: https://annovar.openbioinformatics.org/en/latest/user-guide/download/

Step 2: Fetch Uniprot Annotation with gather_uniprot_annotation.py

2.1. Use the step_2_gather_uniprot_annotation.py script with a local TSV file containing all human genes from Uniprot to annotate the VCF. This will add GOs, BRENDA, and PFAM information to each line of the VCF file.

Step 3: AlphaMissense

3.1. The final script (step_3_match_alphamiss.py) is designed to find the variants in the AlphaMissense table, which contains pre-calculated human proteome variants. Pathogenicity information is added to the final DataFrame. It is recommended to use a table without commented lines or remarks.

The AlphaMissense table can be downloaded from: https://github.com/google-deepmind/alphamissense Step 4: Creating Structures (Future Updates)

4.1. In future updates, the tool will use the AlphaFold database and the pyRosetta molecular modeling suite to generate structures for all variants.