JAX-CNV: A Whole Genome Sequencing-based Algorithm for Copy Number Detection at Clinical Grade Level.

Wan-Ping Lee, The Jackson Laboratory
Qihui Zhu, The Jackson Laboratory
Xiaofei Yang
Silvia Liu
Eliza Cerveira, The Jackson Laboratory
Mallory Ryan, The Jackson Laboratory
Adam Mil-Homens
Lauren Bellfy
Kai Ye
Charles Lee, The Jackson Laboratory
Chengsheng Zhang, The Jackson Laboratory


We aimed to develop a whole-genome sequencing (WGS)-based copy number variant (CNV) calling algorithm with the potential of replacing chromosomal microarray assay (CMA) for clinical diagnosis. JAX-CNV is thus developed for CNV detection from WGS. The performance of this CNV calling algorithm was evaluated in a blinded manner on 31 samples and compared to the results of clinically validated CMAs. Comparing with the 112 CNVs reported by clinically validated CMAs for these 31 samples, JAX-CNV recalled 100% of these CNVs. Besides, JAX-CNV identified an average of 30 CNVs per individual that is an approximately seven-fold increase compared to calls of clinically validated CMAs. Experimental validation of 24 randomly selected CNVs, showed one false positive (i.e., a false discovery rate of 4.17%). A robustness test on lower-coverage data revealed a 100% sensitivity for CNVs greater than 300 kb (the current threshold for College of American Pathologists) down to 10× coverage. For CNVs greater than 50 kb, sensitivities were 100% for coverages deeper than 20×, 97% for 15×, and 95% for 10×. We developed a WGS-based CNV pipeline, including this newly developed CNV caller JAX-CNV, and found it capable of detecting CMA-reported CNVs at 100% sensitivity with about 4% false discovery rate. We propose that JAX-CNV could be further examined in a multi-institutional study to justify the transition of first-tier genetic testing from CMAs to WGS. JAX-CNV is available on https://github.com/TheJacksonLaboratory/JAX-CNV.