Summarizing specific profiles in Illumina sequencing from whole-genome amplified DNA.
Bottom Line: Detailed analysis of the reads from amplified libraries revealed characteristics suggesting that majority of amplified fragment ends are identical but inverted versions of each other.Read coverage in amplified libraries is correlated with both tandem and inverted repeat content, while GC content only influences sequencing in long-insert libraries.To utilize the full potential of WGA to reveal the real biological interest, this article highlights the importance of recognizing additional sources of errors from amplified sequence reads and discusses the potential implications in downstream analyses.
Affiliation: Parasite Genomics, Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Cambridge CB10 1SA, UK Faculty of Medicine, Division of Parasitology, Department of Infectious Disease, University of Miyazaki, Miyazaki 889-1692, Japan.Show MeSH
Related in: MedlinePlus
Mentions: One of the main concerns in using amplification methods is their tendency to form chimeric DNA fragments, which seems to arise from a mechanism involving priming from displaced 3′ termini.17 Based on this mechanism, the majority of chimeric DNA fragments will be sequences where a segment, which is partially deleted, illustrated by Segments a and b in Fig. 1A, joins to another Sequence c from the same chromosome which is inverted. Thus we could measure the tendency of amplified libraries forming chimeras by counting the number of read pairs that were mapped with both mates in the wrong orientation (i.e. forward–forward or reverse–reverse). All amplified libraries show an increased proportion of reads in the wrong orientation when compared with the unamplified counterparts (Table 2). This pattern is more prevalent in long-insert reads, where long-insert libraries prepared using Phi and Tre show 10–12.9% of reads with the wrong orientation compared with 0.55–4.1% in short-insert libraries. The majority of these reads are evenly distributed across the chromosomes in all amplified samples (Supplementary Fig. S3), suggesting that wrongly amplified fragments occur infrequently and randomly. As mentioned before, another characteristic of the chimeras is that part of the sequence will be deleted, and as a result the two segments where they map in the genome will appear to have been brought closer together as a consequence of the deletion (Fig. 1B). Hence, when insert size is calculated based on the mapping positions, we expect to see a much broader insert size distribution than of non-chimeric reads. Indeed, this is what we observed in all WGA long-insert libraries (Replicate 1 of Phi shown in Fig. 1C and rest in Supplementary Fig. S4) with distances between mates mapped in wrong orientation sometimes even >10 kb. By further looking at sequence reads that can be uniquely mapped into two different positions on the same chromosome, we found two cases that confirm the presence of chimeras in Phi and Tre amplified fragments (Supplementary Fig. S5).Figure 1.
Affiliation: Parasite Genomics, Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Cambridge CB10 1SA, UK Faculty of Medicine, Division of Parasitology, Department of Infectious Disease, University of Miyazaki, Miyazaki 889-1692, Japan.