PolyGenius

Guide

Learn PolyGenius step by step — from applying a published score to running a full cross-cohort analysis.

01

Getting started

Why PolyGenius, the mental model, and one complete analysis

  1. 1Why PolyGeniusWhat PolyGenius is for, and how it changes a PGS project
  2. 2Your First AnalysisThe mental model behind PolyGenius, and one complete analysis from import to plot
02

Applying a PGS to cohort data

Getting a model in, a PolyGenius Study, and what compute derives

  1. 1Getting a Model In and OutImporting published scores from the PGS Catalog or a file, and writing your own back out
  2. 2A PolyGenius StudyPolyGeniusStudy: what it holds, how samples are identified, and what it validates
  3. 3Computing Scores and CovariatesScores, relatedness, population structure and the other quantities compute derives
03

Association analyses

Estimating and pooling effects

  1. 1Choosing an Association AnalysisWhat associate does, what comes back, and which analysis answers which question
  2. 2Regression AssociationsOne call, many outcomes: continuous and binary models, subgroups, and comparisons
  3. 3Survival AssociationsTime-to-event outcomes: Cox, competing risks, Kaplan-Meier, and the assumption nobody checks for you
  4. 4MediationSplitting an effect into the part that runs through an intermediate and the part that does not
  5. 5Single-Variant AssociationsA per-variant scan through PLINK2, in the same result shape as everything else
  6. 6Meta-AnalysisCombining association results across cohorts without moving individual-level data
04

Building and prioritizing PGS models

GWAS sources, construction algorithms, and choosing a candidate

  1. 1GWAS SourcesDeclaring summary statistics as inputs: local files, OpenGWAS and the GWAS Catalog
  2. 2Construction AlgorithmsClumping-and-thresholding, LDpred2, lassosum2 and PRS-CS: choosing one and what it costs
  3. 3Running and Debugging a Generation JobScaling a run across sources and algorithms, reading what came back, and what caching does for you
  4. 4Comparing and Choosing ModelsPredictive performance, incremental value, stratification, and how a shortlist is built
05

Visualization

The plotting surface, colour system, and genome tracks

  1. 1PlotsWhat each visualize family draws, what comes back, and how to get a figure out
  2. 2Colour and StylingThe token system, what a palette role claims about your data, and the four levels of control
  3. 3Genome TracksComposing variant, model and library-wide signal on one shared genome axis