Integrated Sequence Assembly-based Approach for Calling Genomic Long Insertion

摘要

With the application and development of high-throughput sequencing technology, the detection methods of structural variants based on sequencing have emerged. However, since the high-throughput sequencing reads is relatively short compared to the previous sequencing reads, it is difficult to detect long insertion. Although assembly-based approach can solve long insertion, the computational resources used for assembly are too complex, resulting in poor results of assembly and final detection. To this end, ISALins was proposed, firstly the initial results of three different detection tools were merged; then high quality soft-cilpped reads and unmapped reads which is the set of most probable reads containing information of insertion were analyzed and extracted around the initial suspect SV breakpoints; finally these reads were assembled using assembly tool based on De Bruijn Graphs. By experimenting on both simulated and real data, we found that the method was superior to the single tool in detecting precision and sensitivity. Compared with the direct combination of call results of multiple tools, in ensuring detection sensitivity of the premise, ISALins significantly improved the detection accuracy.