Validation of low-coverage whole genome sequencing for detection of copy number aberrations in inherited disorders


Keywords: Copy number aberration; whole genome sequencing; validation

Authors: J. Koskenvuo1, P. Salmenperä1, M. Valori1, I. Scheinin1,2,3, M. Gentile1, S. Myllykangas1;
1Blueprint Genetics, Helsinki, Finland, 2VU University Medical Center, Amsterdam, Netherlands, 3University of Helsinki, Helsinki, Finland.

Abstract: Detection of copy number aberrations in clinical diagnostics has been limited to few genes in MLPA and low resolution in microarrays. We have developed and validated a low-coverage whole genome sequencing assay for genome-wide and high-resolution detection of copy number aberrations (CNAs) from inherited disorders. We used Illumina NextSeq500 sequencing system to generate paired-end 40 base reads. 28 reference samples with 34 confirmed chromosomal aberrations and the golden standard reference sample (NA12878) were applied in the validation. Reads were divided into 5 kb bins and difficult to sequence regions (bins representing 5.8% of the genome) were filtered out. Read counts were corrected for GC content and average mappability of each bin. Segmentation and calling algorithms (ODNAseq, DNAcopy and CGHcall) were applied to detect CNAs. The CNA calls from the reference samples were compared to Affymetrix Genome-Wide Human SNP Array 6.0, G-banded karyotyping analysis and fluorescence in situ hybridization (FISH) results. The assay’s sensitivity to detect >100kb deletions and duplications was 0.97. We utilized a golden standard reference sample containing high-quality copy number variation (CNV) calls of variable sizes to demonstrate the assay’s sensitivity to detect smaller changes. The analytical sensitivity was 0.765 for detecting CNVs of 25-50kb in size and 0.990 for detecting CNVs of over 50kb in size. The smallest detected deletion was 10kb. Our results show the validity of the low-coverage whole genome sequencing assay for diagnostic analysis of CNAs in inherited disorders.


View the poster

Like this: