Single Cell DNA Sequencing Simulator

Abstract

Single Cell DNA (scDNA) sequencing is a burgeoning field that allows scientists to sequence the genetic code of one single cell. This process is a departure from bulk sequencing, wherein the genomes of multiple cells are sequenced. Yet current simulators do not account for the critical difference between single cell and bulk sequencing: whole genome amplification (WGA). In this paper, we discuss the two separate steps involved in scDNA sequencing: WGA and the use of next generation sequencing (NGS) technology to sequence the amplified reads, focusing on the former. We analyze the different types of errors that result from WGA and the current models to simulate scDNA sequencing. Then, we present a new model for simulating scDNA sequencing using degenerate-nucleotide-primer PCR (DOP-PCR). Our model generalizes two existing models of PCR and uses a Bienayme-GaltonWatson branching process to model DOP-PCR. Our model uniquely accounts for drop-out, which is the largest source of error in WGA and scDNA sequencing. We evaluate our model after both the DOP-PCR WGA simulation and the NGS simulation, which we do using ART’s Illumina simulator. Our results were evaluated on three metrics: coverage, frequency distribution of reads, and autocorrelation. The first evaluation of the model, after WGA simulation, showed that our model follows the general trends of DOP-PCR, and properly accounts for drop-out. The second evaluation, after NGS simulation, also revealed that our model produces the same general trends as th empirical data, but conclusions on where these trends stemmed from are hard to parse out. Ultimately, we provide a novel, foundational scDNA DOP-PCR simulator that successfully models the amplification error of drop-out.