Supplementary information: Lacking alignments? The next-generation sequencing mapper segemehl revisited

Each artificial dataset consists of 100 000 singleor paired-end reads and was simulated using Mason v0.1.1 [1] from the Human genome (hg19), excluding haplotypes, random contigs, and ‘non-chromosomal’ sequences. For the single-end Illumina datasets, Mason was run in Illumina mode with parameters -hn2, -sq, and the read length (-n) set to 100 and 30 for long and short reads, respectively. For the paired-end dataset, additionally, the parameters -mp, -ll 375, and -le 100 were specified and the read length (-n) was again set to 100. The artificial 454 dataset was simulated in 454 mode with the parameter -hn 2, -sq, -k 0.3, -bm 0.4, -bs 0.2, and -nm 400 (analogously to Langmead & Salzberg [2]). The real datasets were downloaded, converted to fastQ format, some of them post-processed, and downsampled to 100 000 singleor paired-end reads. The Illumina DNA-seq dataset was used as both single-end (by only using the first read sequences) and paired-end data. For both Illumina mRNA-seq datasets, the postprocessing involved removing reads that possibly overlapped exon-exon junctions. To achieve this, the entire dataset was mapped using segemehl (with -S option), STAR [3], TopHat2 [4], and SOAPsplice [5], and reads which were split-mapped by any of these tools were removed prior to down-sampling. In case of the paired-end mRNA-seq dataset, only paired-end reads were kept where both ends were not split-mapped by any of the tools. For Illumina shortRNA-seq, 3-adapter contaminations on the read sequences were clipped using fastx clipper (part of the FASTX-Toolkit) with the Illumina shortRNA-seq adapter (TCGTATGCCGTCTTCTGCTTGT). Before down-sampling, reads outside of the expected length range (19-25nt) were discarded. An overview of the benchmarking datasets, their sequencing platforms, library types, and average read lengths is given in Supplementary Table S1. To permit full reproducability, we have assembled an Electronic Supplement comprising all data, custom scripts, and detailed information on how to re-run the benchmarks.