PGS Catalog is An open database of polygenic scores and the relevant metadata required for accurate application and evaluation.
A polygenic score (PGS) aggregates the effects of many genetic variants into a single number which predicts genetic predisposition for a phenotype.
Repo has a set of notebooks and utility functions for calculating polygenic scores for a genome from raw reads.
It includes notebooks on:
- how to run sequence alignment
- how to run variant calling
- how to pick alternate contigs
- how to annotate variants
- how to filter variants [TODO]
- how to calculate individual's polygenic scores quickly
PGS Catalog doesn't have any special API keys, anyone can query it. So there isn't really anything except a raw human genome file (fastq) you need before starting (vcf file would allow to skip first 2-4 notebooks).
Disclaimer: Genome analysis is computation heavy, some steps here might take a whole day to run depending on the hardware (ex: sequence alignment is especially heavy).
TODO:
- how to search the dna for a specific variance
- polygenic score interpretation. There are many different ways methods how to calculate PGS and score numbers results vary a lot.
# create a virtual environment with your favorite venv tool
sudo apt-get install build-essential python3-dev libsnappy-dev # for pandas to_parquet and read_parquet
pip install -r requirements.txt
pre-commit install