Home

Overview of PolyFun and PolyPred

The PolyFun software can perform several tasks:

1. Computing prior causal probabilities based on functional annotations.
2. Functionally-informed fine-mapping, using either SuSiE or FINEMAP.
3. Leveraging fine-mapping to improve trans-ethnic polygenic risk scores, using PolyPred.
4. In addition, the PolyFun software provides a fully-functioning version of stratified of LD-score regression (adapted to python3), which you can use to estimate functional enrichment of functional annotations.

You can find detailed instructions for each of these tasks on the relevant wiki page (using the menu on the right hand side). Below we provide a general overview of a recommended PolyFun workflow:

Typical PolyFun workflow

A typical PolyFun workflow consists of the following steps:

1. Munging your sumstats to the PolyFun format, using the script munge_polyfun_sumstats.py
2. Computing SNP-specific prior causal probabilities, using the script polyfun.py
3. Fine-mapping (i.e., computing posterior causal probabilities), using the script finemapper.py
4. (optional): Computing polygenic risk scores using the script polypred.py
5. (optional): Estimating polygenic localization using the script polyloc.py

Please note that the software is flexible and does not necessarily adhere to this workflow. For example, you can skip step 2 and then use finemapper.py to perform non-functionally-informed fine-mapping. Please find more details in the wiki pages for specific tasks.