Biases in the SMART-DNA library preparation method associated with genomic poly dA/dT sequences

Abstract

Avoiding biases in next generation sequencing (NGS) library preparation is crucial for obtaining reliable sequencing data. Recently, a new library preparation method has been introduced which has eliminated the need for the ligation step. This method, termed SMART (switching mechanism at the 5′ end of the RNA transcript), is based on template switching reverse transcription. To date, there has been no systematic analysis of the additional biases introduced by this method. We analysed the genomic distribution of sequenced reads prepared from genomic DNA using the SMART methodology and found a strong bias toward long (≥12bp) poly dA/dT containing genomic loci. This bias is unique to the SMART-based library preparation and does not appear when libraries are prepared with conventional ligation based methods. Although this bias is obvious only when performing paired end sequencing, it affects single end sequenced samples as well. Our analysis demonstrates that sequenced reads originating from SMART-DNA libraries are heavily skewed toward genomic poly dA/dT tracts. This bias needs to be considered when deciding to use SMART based technology for library preparation.

Facile single-stranded DNA sequencing of human plasma DNA via thermostable group II intron reverse transcriptase template switching

“…After completion of DNA synthesis, the RT template switches to an acceptor DNA oligonucleotide thereby adding a second primer-binding site for PCR and DNA-seq adapter addition. This method is relatively rapid, but poly(dT) tailing and variable non-templated nucleotide addition by the retroviral RT to the 3′ end of the DNA product prior to template switching make it difficult to precisely identify the DNA template ends and can introduce strong bias for regions with long poly(dA) or poly(dT) runs 23 . A recently developed commercial kit (ACCEL-NGS 1 S PLUS, Swift Biosciences) with a processing time of 2 h up to the PCR amplification step utilizes a proprietary adaptase technology to repair, tail, and attach a DNA-seq adapter to the 3′ end of a DNA strand.…”

SMARTcleaner: identify and clean off-target signals in SMART ChIP-seq analysis

“…1a). As mentioned previously [14], the SMART poly(dA) primers can anneal to poly(T) sequences that are either located within the IP-DNA fragments (Fig. 1b) or present in non-target DNA fragments (i.e., the DNA fragments pulled down during IP non-specifically) (Fig.…”

“…After amplification, sequencing, and read mapping (note that only one strand of the dsDNA is sequenced), ChIP-seq reads from poly(T/A) genomic DNAs, due to false priming and amplification, will accumulate next to the poly(T/A) sites in a clear strand-specific manner because the poly(dA) primers only anneal to the DNA strand containing poly(T). To illustrate this, we examined the reads in a human ChIP-seq sample (Additional file 1: Table S1, Dataset 1, SRR3229031) that was prepared using the Clontech DNA SMART ChIP-seq kit and by PE sequencing [14]. As this particular dataset was obtained from sequencing of control samples (i.e., input DNA), no genomic regions would be expected to show ChIP-seq read enrichment.…”

“…It is conceivable that the poly(dA) primer, however, can lead to signals amplified from non-targeted genomic regions containing consecutive Ts. Indeed, a previous study of SMART ChIP-seq reads revealed a strong bias of base constitution at the 3′ end of the sequenced reads that are enriched near long (≥12 bp) poly(T/A) containing genomic loci [14]. The authors proposed a computational strategy to reduce this bias by normalizing the ChIP-seq data for the genomic abundance of different polyN tracts, but only achieved partial success [14].…”

“…Indeed, a previous study of SMART ChIP-seq reads revealed a strong bias of base constitution at the 3′ end of the sequenced reads that are enriched near long (≥12 bp) poly(T/A) containing genomic loci [14]. The authors proposed a computational strategy to reduce this bias by normalizing the ChIP-seq data for the genomic abundance of different polyN tracts, but only achieved partial success [14]. Here, we revisited this problem and demonstrated that the unique features of the falsely amplified reads can be exploited to effectively remove artifact ChIP-seq reads from SMART protocols.…”