NPBSS: a new PacBio sequencing simulator for generating the continuous long reads with an empirical model

Abstract

BackgroundPacBio sequencing platform offers longer read lengths than the second-generation sequencing technologies. It has revolutionized de novo genome assembly and enabled the automated reconstruction of reference-quality genomes. Due to its extremely wide range of application areas, fast sequencing simulation systems with high fidelity are in great demand to facilitate the development and comparison of subsequent analysis tools. Although there are several available simulators (e.g., PBSIM, SimLoRD and FASTQSim) that target the specific generation of PacBio libraries, the error rate of simulated sequences is not well matched to the quality value of raw PacBio datasets, especially for PacBio’s continuous long reads (CLR).ResultsBy analyzing the characteristic features of CLR data from PacBio SMRT (single molecule real time) sequencing, we developed a new PacBio sequencing simulator (called NPBSS) for producing CLR reads. NPBSS simulator firstly samples the read sequences according to the read length logarithmic normal distribution, and choses different base quality values with different proportions. Then, NPBSS computes the overall error probability of each base in the read sequence with an empirical model, and calculates the deletion, substitution and insertion probabilities with the overall error probability to generate the PacBio CLR reads. Alignment results demonstrate that NPBSS fits the error rate of the PacBio CLR reads better than PBSIM and FASTQSim. In addition, the assembly results also show that simulated sequences of NPBSS are more like real PacBio CLR data.ConclusionNPBSS simulator is convenient to use with efficient computation and flexible parameters setting. Its generating PacBio CLR reads are more like real PacBio datasets.Electronic supplementary materialThe online version of this article (10.1186/s12859-018-2208-0) contains supplementary material, which is available to authorized users.

smsMap: mapping single molecule sequencing reads by locating the alignment starting positions

“…The E. coli MG1655 genome sequence (with the length of 4,614,652 bp) from NCBI (No. NC_000913.3) was downloaded and inputted to the NPBSS simulator [ 19 ] for generating the PacBio simulated reads with different error rates. As a result, 6 simulated datasets with 5, 10, 15, 20 25 and 30% error rates were generated.…”

“…In order to estimate the capability of smsMap for mapping reads that span structural variations (SVs), we used another simulation dataset from chr1 of NA12878 with SVs. The simulation dataset with SVs was generated by inserting 7 SVs (i.e., 3 insertions, 3 deletions and 1 inversion) from DGV [ 38 ] into the reference chr1 and using the NPBSS simulator [ 19 ] at 20x coverage. Among the simulated reads, a total of 185 reads cover the SVs breakpoints.…”

“…All existing tools for mapping short reads, such as BLAST [ 11 ], BLAT [ 10 ], SOAP2 [ 12 ], Bowtie [ 13 ] and BWA [ 14 ], aim to find a “long” substring that would exactly match its mapping locus on the reference genome by using either Burrows-Wheeler Transform Full-text Minute-space (BWT-FM) index [ 15 , 16 ], or substring hashing [ 17 ], or hybrid methods of combining FM index with hashing [ 18 ]. However, due to a higher number of errors which are primarily insertions and deletions rather than substitutions, the mapping methods created for short reads are not readily extended to long reads generated from the single molecule sequencing (SMS) technology [ 19 ].…”

Twelve quick steps for genome assembly and annotation in the classroom

“…No matter which approach is taken, the essential part is to have firsthand experience to select proper computational design and pipeline and to accurately interpret analyzed genome data. Due to its extensive range of analytical tools and application areas, employing an effective simulator (from the quality of raw reads to assembly evaluation) has become an essential step for benchmarking genomic and bioinformatics analyses [92][93][94]. In simulations, considering a (very) large number of datasets is generally not a problem, except when the analysis of each dataset is hugely computationally expensive (e.g., in the genome assembly stage).…”

Technology and Species Independent Simulation of Sequencing Data and Genomic Variants

“…Simulators tailored on Pacific Bioscience SMRT sequencing are more common. In particular, the SimLoRD [17] can only simulate circular consensus sequence reads (CCS) while PaSS [18] and NPBSS [19] can also produce continuous long reads (CLR). Due to its complexity and rapid evolution, the Oxford Nanopore Sequencing simulation is more challenging and hence less common.…”