Performance difference of graph-based and alignment-based hybrid error correction methods for error-prone long reads

The error-prone Third Generation Sequencing (TGS) long reads can be corrected bythe high quality Second Generation Sequencing (SGS) short reads, which is referredas hybrid error correction. We here investigate the influences of the principalalgorithmic factors of two major types of hybrid error correction methods bymathematical modelling and analysis on both simulated and real data. Our studyreveals the distribution of accuracy gain with respect to the original long read error rate.Compared with the alignment-based method, the graph-based method has remarkableadvantage on correcting the long reads with high error rates, while the latter performsbetter on correcting the long reads with relatively shorter lengths. We also demonstratethat the original error rate 19% is the limit for perfect correction, beyond which longreads are too error-prone to be corrected by these methods.