Performance of SHERA, FLASH, COPE, and ARF-PE on merging paired reads.

The four tools are applied to merge paired reads simulated from the R. sphaeroides genome at various error rates. Following the definition in the FLASH study, a merged read is correct when its length equals the corresponding fragment length. Note that the two categories “correct non-merge” and “incorrect non-merge” in the FLASH study are combined together. Because ARF-PE can merge paired reads with little or no overlap, the two categories do not apply for ARF-PE. For each error rate, the best results are shown in bold.