Prioritization of genes driving congenital phenotypes of patients with de novo genomic structural variants

• Main Authors: Sjors Middelkamp, Judith M. Vlaar, Jacques Giltay, Jerome Korzelius, Nicolle Besselink, Sander Boymans, Roel Janssen, Lisanne de la Fonteijne, Ellen van Binsbergen, Markus J. van Roosmalen, Ron Hochstenbach, Daniela Giachino, Michael E. Talkowski, Wigard P. Kloosterman, Edwin Cuppen
• Format: Article
• Language: English
• Published: BMC 2019-12-01
• Series: Genome Medicine
• Subjects: Structural variation; Copy number variants; Neurodevelopmental disorders; Intellectual disability; Multiple congenital anomalies; Driver genes
• Online Access: https://doi.org/10.1186/s13073-019-0692-0

Abstract Background Genomic structural variants (SVs) can affect many genes and regulatory elements. Therefore, the molecular mechanisms driving the phenotypes of patients carrying de novo SVs are frequently unknown. Methods We applied a combination of systematic experimental and bioinformatic methods to improve the molecular diagnosis of 39 patients with multiple congenital abnormalities and/or intellectual disability harboring apparent de novo SVs, most with an inconclusive diagnosis after regular genetic testing. Results In 7 of these cases (18%), whole-genome sequencing analysis revealed disease-relevant complexities of the SVs missed in routine microarray-based analyses. We developed a computational tool to predict the effects on genes directly affected by SVs and on genes indirectly affected likely due to the changes in chromatin organization and impact on regulatory mechanisms. By combining these functional predictions with extensive phenotype information, candidate driver genes were identified in 16/39 (41%) patients. In 8 cases, evidence was found for the involvement of multiple candidate drivers contributing to different parts of the phenotypes. Subsequently, we applied this computational method to two cohorts containing a total of 379 patients with previously detected and classified de novo SVs and identified candidate driver genes in 189 cases (50%), including 40 cases whose SVs were previously not classified as pathogenic. Pathogenic position effects were predicted in 28% of all studied cases with balanced SVs and in 11% of the cases with copy number variants. Conclusions These results demonstrate an integrated computational and experimental approach to predict driver genes based on analyses of WGS data with phenotype association and chromatin organization datasets. These analyses nominate new pathogenic loci and have strong potential to improve the molecular diagnosis of patients with de novo SVs.