Experimental and Computational Approaches to Identify Noncoding Pathogenic Variation in Rare Disease

Noncoding variants occur within noncoding genes as well as within the regulatory nontranslated regions of protein-coding genes. It is important to be aware that these variants have been increasingly implicated in developmental disease through a variety of mechanisms. However, they remain difficult to interpret clinically due to their unclear effect on transcript or protein abundance compared with coding variants. Here, we review methods to identify pathogenic noncoding variants in rare disease, which can present challenges due to the inaccessibility of disease-relevant tissue for many conditions. We explore experimental approaches such as high-throughput functional assays, omic data integration, and long-read sequencing. We also review computational methods for annotating and filtering variants, as well as machine learning methods for predicting variant effect and pathogenicity. We discuss the recent discovery of several developmental syndromes caused by noncoding variants and propose an integrated approach to identifying pathogenic noncoding variants within this patient cohort.