GWAS Workflow Overview

Study Design

• Study Type Selection (Case-control, Cohort, Population-based)
• Sample Size Calculation & Power Analysis
• Inclusion/Exclusion Criteria Definition
• Phenotype Definition & Diagnostic Criteria
• Covariate Strategy Planning
• Ethical Approval & Protocol Development

Planning the study design before data collection: selecting the appropriate study type based on research questions, calculating required sample sizes for adequate statistical power, defining clear inclusion and exclusion criteria, establishing standardized phenotype definitions and diagnostic criteria, planning covariate strategies, and obtaining ethical approval and developing study protocols.

Sample Collection

• Cohort Recruitment (Disease-oriented, Population-based, Case-control)
• Informed Consent & Ethical Approval
• Biological Sample Collection (DNA, Serum, etc.)
• Clinical Information Collection (Phenotypes, Demographics)
• Sample Storage & Management

Recruiting participants, collecting biological samples (DNA, serum) and clinical information from patients or population cohorts. Example: BioBank Japan (BBJ) collects samples from 270,000 patients across 51 diseases through cooperative medical institutions nationwide.

Sequencing / Genotyping (WGS, WES, SNP Arrays)

Determining genetic variants using whole genome sequencing (WGS), whole exome sequencing (WES), or genotyping arrays. SNP arrays are cost-effective for large-scale studies (500K-5M variants), while WGS provides complete genome coverage for rare variant discovery.

Raw Data Generation (FASTQ, BAM, VCF, Genotype Calls)

• Format Conversion (FASTQ, BAM, VCF, Genotype Calls)
• Data Harmonization (Multi-platform, Multi-cohort Standardization)

Converting sequencing reads or array data into standardized formats: FASTQ (raw reads), BAM (aligned sequences), VCF (variant calls), or genotype files (PLINK, BED/BIM/FAM formats) for downstream analysis. Harmonizing data across different platforms, arrays, or cohorts to ensure consistent variant calling, coordinate systems (GRCh37/GRCh38), and allele coding for multi-cohort or multi-platform studies.

Data Quality Control (QC)

• Sample QC (Missingness, Sex Check, Heterozygosity)
• Variant QC (MAF, HWE, Call Rate)
• Relatedness Check (Duplicate Detection, IBD)
• Population Structure (PCA)

Filtering low-quality samples and variants: removing samples with high missingness or sex mismatches, filtering variants by minor allele frequency (MAF) and Hardy-Weinberg equilibrium (HWE), detecting duplicates and related individuals, and assessing population structure using principal component analysis (PCA).

Pre-GWAS Processing

• Phasing (Haplotype Reconstruction)
• Imputation (Infer Ungenotyped Variants)
• Population Structure (PCA, Ancestry)
• Covariate Preparation

Preparing data for association testing: phasing haplotypes (determining parental chromosome origin), imputing ungenotyped variants using reference panels (1000 Genomes, TOPMed), correcting for population structure, and preparing covariates (age, sex, principal components).

Phenotype Preparation

• Phenotype QC (Outlier Detection, Range Validation)
• Normalization/Transformation (INT, Log Transform, Z-score)
• Covariate Adjustment (Age, Sex, Medication Effects)
• Case/Control Definition (Binary Traits)
• Missingness Handling

Preparing phenotype data for association testing: performing quality control to detect outliers and validate ranges, applying appropriate transformations (rank-based inverse normal transformation for quantitative traits, log transforms for skewed data), adjusting for covariates and medication effects, defining cases and controls for binary traits, and handling missing phenotype data appropriately.

GWAS Association Testing

• Association Testing (Linear/Logistic Regression, LMM)
• Multiple Testing Correction (Genome-wide Significance)
• Summary Statistics (Effect Sizes, P-values)

Testing associations between each genetic variant and the trait using linear regression (quantitative traits) or logistic regression (binary traits), with linear mixed models (LMM) to account for relatedness. Applying multiple testing correction (genome-wide significance threshold p < 5×10⁻⁸) and generating summary statistics (effect sizes, p-values).

Replication

• Independent Cohort Validation
• Cross-population Replication
• Effect Size Consistency
• Directional Consistency Check

Validating significant associations in independent cohorts or populations to confirm findings and reduce false positives. Replication studies test whether associations discovered in the discovery cohort can be reproduced in different samples, often with more stringent significance thresholds. Successful replication strengthens confidence in the association and its generalizability across populations.

Post-GWAS Analysis

• Variant Annotation (Functional Consequences)
• Heritability Estimation (SNP Heritability)
• Fine-mapping (Causal Variant Identification)
• Gene/Pathway Analysis (MAGMA, Enrichment)
• Colocalization (eQTL, pQTL Integration)
• Polygenic Risk Scores (PRS Construction)
• Mendelian Randomization (Causal Inference)
• Meta-analysis (Multi-cohort Integration)
• Genetic Correlation (Cross-trait Analysis)

Interpreting GWAS results: annotating variants with functional consequences, estimating heritability, fine-mapping causal variants, analyzing gene and pathway enrichment, integrating with molecular QTL data (eQTL, pQTL), constructing polygenic risk scores, performing Mendelian randomization for causal inference, meta-analyzing across cohorts, and estimating genetic correlations between traits.

Functional Validation

• Functional Genomics (eQTL, CRISPR, Epigenomics)
• Cell Culture Models (In Vitro Validation)
• Animal Models (In Vivo Studies)
• Drug Target Validation (Therapeutic Development)

Experimental validation of GWAS findings: functional genomics studies (eQTL analysis, CRISPR editing, epigenomic profiling), in vitro validation using cell culture models, in vivo studies using animal models, and drug target validation for therapeutic development.