Analysis of structural variants in osteoporosis

Abstract

Osteoporosis is a metabolic bone disorder with a strong genetic influence which has been a subject of study in research aiming to pinpoint crucial genes associated with bone health. This underlying pathophysiology has been linked to both common and rare genetic variations, and to a lesser extent, genomic structural variants (SVs), which are defined as alterations in chromosome structure exceeding 50 base pairs (bp) in size. Despite advances in sequencing, SVs in osteoporosis remain difficult to reliably detect due to the short read length (<300 bp) of 2nd generation sequencing, the diverse spectrum of SVs, and ongoing challenges in achieving high sensitivity and specificity in variant identification. The aim of this study was to conduct a comprehensive analysis of specifically selected SV detection tools using a 2-generation Maltese family having multiple relatives affected with osteoporosis and low bone mineral density (BMD). To achieve this objective, BreakDancer, Pindel, and Lumpy were carefully selected and computationally assessed on six readily available BAM files generated from short-read whole-genome sequencing (WGS). Genotype calling was required for BreakDancer and Lumpy, and additional tools BreakDown and SVtyper were employed. Variants identified by these tools were further annotated using the Variant Effect Predictor (VEP) to assess their impact on genes, transcripts, protein sequences, and regulatory regions. A number of filtering steps were performed to narrow down the list of variants and prioritise SVs located in genes with relevant to bone physiology. Ten shortlisted SVs underwent a computational visualisation using Integrative Genomics Viewer (IGV) and experimental validation by PCR sizing and Sanger sequencing. Among these, three were experimentally confirmed, derived from the Lumpy output: ARHGEF3 g.57003274_57003433del, TBX15 g.119482201_119483619del, and ADAM9 g.38953621_38953881del. Likewise, two others were experimentally confirmed from the Pindel output: SOD2 g.160086257_160088689del and KLF12 g.74284677_74284844dup,. Overall, both Lumpy and Pindel demonstrated effectiveness in detecting SVs, with 5 out of 10 SVs identified as true positives. Around 400 SVs were called by all three tools. No variants were shortlisted by BreakDancer. Lumpy exhibited superiority over Pindel and BreakDancer, showcasing faster runtime, smaller memory footprint for output files, and minimal system requirements. In conclusion, the findings suggest that the SVs detected by Lumpy and Pindel could potentially be contributing to the genetic architecture of osteoporosis and BMD.