Fig 1.
Upper panel (a) shows the results for POLCA, Pilon, Racon, and NextPolish in correcting simulated errors for five different experiments with different numbers of errors introduced into an assembly of the Arabidopsis thaliana genome. Lower panel (b) shows the running times (wall clock time) of each program, measured on a 16-core AMD Opteron system with 128Gb of RAM, running with 16 threads. The run times do not include the time spent on mapping the reads, which was the same for all programs.
Table 1.
Results for error correction by POLCA and Pilon on an A. thaliana genome (total size 119Mb) with three different numbers of simulated errors.
Error rates ranged from 0.1% to 0.46%. Boldface indicates the better values for each experiment in each row.
Table 2.
Polishing results for four Klebsiella pneumoniae isolates.
The columns list average identity rates for 1-to-1 best alignments of the polished assemblies to the finished sequences of the isolates. In bold we highlight the best result and any result within 0.01% of the best.
Table 3.
Total number of erroneous bases (lower is better) remaining in the Arabidopsis thaliana genome with 231,929 introduced errors after polishing by two methods run consecutively.
The program shown in each row was run first, followed by the program shown in each column. The “Single run” column shows the number of errors remaining after a single run of each program. Note that in some cases the total number of errors increases after running two programs consecutively, such as after using Pilon or Racon on assemblies polished with NextPolish or POLCA.