Luo, Can; Liu, Yichen Henry; Liu, Han; Zhang, Zhenmiao; Zhang, Lu; Peters, Brock A.; Zhou, Xin Maizie. (2026).听.听Bioinformatics Advances, 6(1), vbag267.听
Accurately identifying genetic differences is important for understanding human health and disease. Standard short-read DNA sequencing can reliably detect small genetic changes but often struggles with larger structural variants (SVs), especially in complex regions of the genome. Linked-read sequencing helps address this limitation by adding molecular barcodes that connect DNA fragments originating from the same longer molecule. In this study, we developed stLFR-sim, an open-source simulator, to test whether longer barcoded reads could improve SV detection. Simulations based on a well-characterized human genome showed that increasing read length consistently improved accuracy, with 1000-base-pair reads performing best among the tested configurations and approaching the performance of long-read sequencing methods. These findings demonstrate the potential of longer barcoded reads for improving structural variant detection and provide a simulation tool for evaluating future sequencing technologies.
Figure 1
Simulated and real PE100_stLFR SV calling performance comparison聽Truvari聽(v4.0.0) was applied to benchmark the SV callset against the GIAB SV benchmark set. Across 12 different simulation configurations, the minimum (Min_sim), maximum (Max_sim) and average (Avg_sim) value of F1, recall and precision score were calculated. (a) F1, recall, precision for insertion SV calling. (b) F1, recall, precision for deletion SV calling.