Simulated Illumina short read datasets used for the evaluation of adapter trimming tools

All data are 100bp paired-end Illumina reads simulated from the human genome (hg19.fa.masked) using pIRS which can simulate reads with realistic sequencing error profile and GC-bias. For paired-end reads simulation, pIRS selects randomly select insert sizes from a normal distribution with preset mean and standard deviation values. To simulate adapter contamination, the pIRS was modified so that when selected insert sizes are shorter than the preset read length (100bp in this study), adapter sequences will be concatenated at ends of genomic sequences before subsequent simulation of sequencing errors. Multiple simulations were performed with mean values of insert size ranging from 80 to 130 in order to approximate high to low levels of adapter contamination (the mean value used for each simulation was included in the file name of simulated dataset). All simulations were done with standard deviation of 10bp and 0.1x coverage.