CNV Analysis Using Illumina Global Screening Array (GSA)

This document explains how to perform and interpret Copy Number Variation (CNV) analysis using GenomeStudio and GSA genotyping arrays, including:

• Role of Manifest Files
• Role of Cluster Files
• Interpretation of Log R Ratio (LRR)
• Interpretation of B Allele Frequency (BAF)
• Common CNV patterns and biological meaning
• Literature references relevant to GSA CNV workflows

---

1. What is a Manifest File?

A Manifest File contains probe and SNP annotation for the Illumina array.

Purpose

• Maps SNP ID, chromosome, genomic position
• Defines A/B alleles and probe sequences
• Required for genotype calling and CNV analysis
• Ensures GenomeStudio knows how to interpret raw array signals

Where to get it

✔ Illumina downloads
✔ Provided during array purchase
✔ Format: .bpm, .csv, or .txt

---

2. What is a Cluster File?

A Cluster File (.egt) contains predefined genotype centroids for AA, AB, BB calls.

Purpose

• Standardizes genotype calling across batches
• Reduces sample-specific shifts in clustering
• Improves consistency in multi-batch or low-quality samples

Feature	Manifest File	Cluster File
Defines	SNP location & annotation	Expected genotype clusters
Needed for	CNV + genotyping	Genotyping only

➡ Note: Cluster file is NOT required for CNV-only intensity analysis

---

3. GenTrain Score

A quality metric (0–1) measuring cluster separation.

GenTrain Score	Interpretation
≥ 0.7	High quality
0.4–0.7	Review required
≤ 0.3	Poor—exclude SNP

Poor GenTrain SNPs degrade LRR/BAF reliability.

---

4. CNV Detection in GenomeStudio

CNV calling relies on two intensity-based metrics:

4.1 Log R Ratio (LRR)

Measures total signal intensity → detects copy number change.

LRR Value	CN State	Interpretation
~0.00	CN = 2	Normal
-0.3 to -0.8	CN = 1	Heterozygous deletion
≤ -1.0	CN = 0	Homozygous deletion
+0.2 to +0.5	CN = 3	Duplication
> +0.6	CN ≥ 4	Amplification

---

4.2 B Allele Frequency (BAF)

Fraction of B-allele intensity.

BAF Pattern	Interpretation
0.0, 0.5, 1.0	Normal diploid
Only 0.0 + 1.0	Loss of heterozygosity / deletion
0.0, 0.33, 0.66, 1.0	Duplication (CN=3)
Random scattered 0–1	Mosaicism or UPD

---

5. Interpretation Workflow

✔ Check LRR → Does it shift up/down?
✔ Check BAF → Are normal clusters preserved?
✔ Compare region to:

• DGV
• ClinVar
• DECIPHER
• Internal CNV database

✔ Apply ACMG-CNV criteria (e.g., ClinGen technical standards)

---

6. Example CNV Interpretation

Pattern	LRR	BAF	Interpretation
Flat at 0.0	0.0	0,0.5,1	Normal
Negative shift	-0.5	0 & 1 only	Heterozygous deletion
Positive shift	+0.35	0,0.33,0.66,1	Duplication

---

7. When NOT to Use Genotype Cluster File

❌ CNV calling does NOT require genotype calls
✔ "Intensity Only" mode can be used
✔ Genotype failures do NOT stop CNV analysis

---

8. Key Literature

Highly recommended papers for interpretation workflow:

1️⃣ The individual and global impact of copy-number variants on complex human traits
📎 https://doi.org/10.1016/j.ajhg.2022.03.013

2️⃣ Development and validation of a pharmacogenomics reporting workflow based on the Illumina GSA array
📎 https://doi.org/10.3389/fphar.2024.1349203

---

9. Notes From Practical Use

✔ GSA reliably detects CNVs > 50–100 kb
✔ Mosaic CNVs require manual inspection
✔ Always verify large CNVs with:

• IGV (BAM-based depth)
• MLPA or qPCR (clinical validation)
• SNP density matters (weak signal in telomeric / centromeric regions)