Abstract
Our results indicate that CNVs detected by arrays may be the coincidental co-localization of smaller CNVs, whose presence is more likely to perturb an aCGH hybridization profile than the effect of an isolated, small, copy number alteration. Our findings help explain the hitherto unexplored discrepancies between array-based studies of copy number variation in the mouse genome.
Introduction
Copy Number Variants (CNVs)—genomic segments that vary in the number of copies between individuals—are fundamental drivers of phenotypic diversity and disease susceptibility. In the mouse model, however, a persistent "concordance gap" has plagued structural variation research. Different high-throughput studies focusing on the same inbred strains frequently report non-overlapping sets of CNVs, raising questions about the reliability of current detection technologies and algorithms.
Methodology
To investigate these discrepancies, we conducted array Comparative Genomic Hybridization (aCGH) using high-density NimbleGen arrays (2.1 million probes) across seven inbred mouse strains (A/J, AKR/J, BALB/cJ, C3H/HeJ, CBA/J, DBA/2J, and LP/J), with C57BL/6J as the reference.
A critical component of our workflow was SNP Standardization. We identified that Single Nucleotide Polymorphisms (SNPs) within probe sequences significantly interfere with hybridization signals; a single SNP can reduce the $log_2$ ratio by approximately 0.5. By correcting for these 8.27 million known SNPs, we minimized false positives stemming from probe-level noise. Validation was performed using a multi-modal approach including PCR, sequencing, real-time PCR, FISH, and MLPA.
The Architecture of Complexity
Our analysis reveals that the primary cause of study discordance is the fundamental molecular structure of CNVs. We distinguish between two types:
- Simple CNVs: Continuous blocks of DNA gain or loss.
- Complex CNVs: Clusters of small, interspersed deletions and gains separated by segments of normal DNA.
Counter-intuitively, arrays are more likely to detect these complex regions because the cumulative signal perturbation from multiple small variations is easier to distinguish from background noise than a single, isolated small variant. However, this "Array Paradox" means that the reported boundaries of these variants are often imprecise, leading to low overlap between different platforms and algorithms.
Key Results and Error Rates
The study identified 1,477 deletions and 499 gains, covering approximately 4.4% of the mouse genome. Functional analysis showed these regions are enriched for genes related to immunity and environmental response (e.g., odorant binding) but depleted for essential housekeeping genes.
Most significantly, molecular validation exposed high error rates in current aCGH-based catalogs:
- False Positive Rate: ~32-35%.
- False Negative Rate: Estimated at >50%.
These figures suggest that a substantial portion of the structural variation landscape remains hidden or misinterpreted by current array technologies. Simple CNVs showed much higher replication rates across studies compared to their complex counterparts.
Conclusion
The "elusive" nature of mouse CNVs—characterized by small sizes and fragmented architectures—places a ceiling on the efficacy of hybridization-based methods. While aCGH has provided a foundational map, the high false negative rate and boundary ambiguity necessitate a transition to Next-Generation Sequencing (NGS). Sequencing through the breakpoints of these complex regions is the only way to resolve the blurred image provided by arrays and achieve a high-resolution map of the mouse genome's structural variation.
Summary updated