Appendix B — Templates

Copy these into your project and fill them in. They are short on purpose: a template nobody fills in protects nothing.

B.1 Analysis plan

Write it before the first association test. It is what lets you, a reviewer, or your future self tell a planned analysis from a fishing trip.

# Analysis plan — <trait> GWAS

- **Date / version:** 2026-01-15 / v1
- **Question:** Which common variants are associated with <trait>?
- **Data:** <biobank>, release <x>, genome build GRCh38, imputed to <panel>
- **Phenotype definition:** cases = <codes / criteria>; controls = <criteria>;
  exclusions = <criteria>
- **Ancestry groups analyzed:** <groups>, each analyzed separately, then meta-analyzed
- **Covariates:** age, sex, PC1–PC10, genotyping batch
- **QC thresholds:** geno 0.02, mind 0.02, MAF 0.01, HWE 1e-6, het ±3 SD,
  KING 0.0884, INFO ≥ 0.8
- **Model and software:** SAIGE <version>, binary trait, LOCO on
- **Significance:** p < 5e-8; suggestive p < 1e-5
- **Planned follow-up:** clumping (r² 0.1, 500 kb), annotation (VEP), eQTL lookup
- **Changes after this date:** recorded below, with reason

B.2 QC report

One table per step makes the attrition easy to audit.

| Step | Samples | Variants | Removed | Reason / threshold |
| --- | --- | --- | --- | --- |
| Raw data | | | | |
| Variant missingness | | | | --geno 0.02 |
| Sample missingness | | | | --mind 0.02 |
| Sex check | | | | F-statistic |
| Heterozygosity | | | | ±3 SD |
| Relatedness | | | | KING > 0.0884 |
| MAF / HWE | | | | MAF 0.01, HWE 1e-6 |
| Final | | | | |

Attach the heterozygosity-versus-missingness plot, the PCA plot, and the QQ plot.

B.3 Methods paragraph

A starting point for the methods section of a paper or report. Replace the brackets; delete what you did not do.

Genotypes were quality controlled with PLINK [version]. We excluded variants with missingness > [2]%, minor allele frequency < [1]%, or Hardy–Weinberg p < [1×10⁻⁶], and samples with missingness > [2]%, heterozygosity beyond ±3 SD of the mean, or discordant genetic and reported sex. Genetic ancestry was inferred by projecting samples onto [reference] principal components. Association testing used SAIGE [version], a generalized linear mixed model that accounts for relatedness and case–control imbalance, adjusting for age, sex, and the first [10] principal components. Genome-wide significance was set at p < 5×10⁻⁸. Results across ancestry groups were combined by inverse-variance-weighted fixed-effects meta-analysis in METAL [version].

B.4 Project folder layout

project/
├── README.md          # question, data source, how to rerun
├── plan.md            # the analysis plan above
├── config/            # thresholds and paths in one place
├── scripts/           # numbered in run order: 01_qc.sh, 02_pca.sh, ...
├── logs/              # scheduler and tool logs
├── results/           # outputs, never edited by hand
└── reports/           # QC report, plots, tables for the paper

Raw data stays where the data provider put it and is read in place. Never copy controlled-access data into a personal folder or a Git repository.