LRCstats is an open-source pipeline for benchmarking DNA long read correction algorithms for long reads outputted by third generation sequencing technology such as machines produced by Pacific Biosciences. The reads produced by third generation sequencing technology, as the name suggests, are longer in length than reads produced by next generation sequencing technologies, such as those produced by Illumina. However, long reads are plagued by high error rates, which can cause issues in downstream analysis. Long read correction algorithms reduce the error rate of long reads either through self-correcting methods or using accurate, short reads outputted by next generation sequencing technologies to correct long reads.
Of course, some long read correction algorithms are better than others, and developers of long read correction algorithms will wish to compare their algorithm with others currently available. LRCstats benchmarks long read correction algorithms using long reads produced by simulators (such as SimLoRD or PBSim) where the two-way alignments between the uncorrected long reads (uLR) and the corresponding sequences in the reference genome (Ref) are given in some sort of alignment file and then aligning the corrected long reads (cLR) to the Ref-uLR two-way alignments to create three-way alignments using a dynamic programming algorithm. Statistics on these three-way alignments are then collected, such as the overall error rates of the corrected long reads.
