Identifying the best PCR enzyme for library amplification in NGS

Abstract

Background. PCR amplification is a necessary step in many next-generation sequencing (NGS) library preparation methods [1, 2]. Whilst many PCR enzymes are developed to amplify single targets efficiently, accurately and with specificity, few are developed to meet the challenges imposed by NGS PCR, namely unbiased amplification of a wide range of different sizes and GC content. As a result PCR amplification during NGS library prep often results in bias toward GC neutral and smaller fragments. As NGS has matured, optimized NGS library prep kits and polymerase formulations have emerged and in this study we have tested a wide selection of available enzymes for both short-read Illumina library preparation and long fragment amplification ahead of long-read sequencing.We tested over 20 different hi-fidelity PCR enzymes/NGS amplification mixes on a range of Illumina library templates of varying GC content and composition, and find that both yield and genome coverage uniformity characteristics of the commercially available enzymes varied dramatically. Three enzymes Quantabio RepliQa Hifi Toughmix, Watchmaker Library Amplification Hot Start Master Mix (2X) 'Equinox' and Takara Ex Premier were found to give a consistent performance, over all genomes, that mirrored closely that observed for PCR-free datasets. We also test a range of enzymes for long-read sequencing by amplifying size fractionated S. cerevisiae DNA of average size 21.6 and 13.4 kb, respectively.The enzymes of choice for short-read (Illumina) library fragment amplification are Quantabio RepliQa Hifi Toughmix, Watchmaker Library Amplification Hot Start Master Mix (2X) 'Equinox' and Takara Ex Premier, with RepliQa also being the best performing enzyme from the enzymes tested for long fragment amplification prior to long-read sequencing.

Keywords: PCR; amplification; barcode; illumina; indexing; next-generation sequencing; polymerase.

Fig. 1.. Low coverage index (fraction of genome covered at <50 % mean coverage) obtained after 14 cycles of amplification with 1 ng of each test microbial genome; BP: Bordetella pertussis, EC: Escherichia coli, CD: Clostridioides difficile, and PF: Plasmodium falciparum. For each PF was also amplified using a 94 °C denaturation temperature, ‘PF 94C’.

Fig. 2.. Ranked low coverage index (fraction of genome covered at <50 % mean coverage) values obtained after 14 cycles of amplification with 1 ng of each test microbial genome; (a) Bordetella pertussis, (b) Escherichia coli, (c) Clostridioides difficile, and (d) Plasmodium falciparum.

Fig. 3.. Multi-genome average low coverage index (fraction of genome covered at <50 % mean coverage) ranking taking the sum of the LCI values for all four genomes compared to that obtained from PCR free data.

Fig. 4.. Screenshot from Artemis genome browser [7] over regions of (a) B. pertussis and (b) P. falciparum genomes. In each the top panel is a GC content plot with the maxima and minima GC content of each region displayed at the far right. The bottom panel is an open read frame plot and the middle panel plots read coverage over each base. The inset panel on each denotes the enzyme used for each coloured line. In (a) RepliQa (pink), Watchmaker (blue) and PCR-free data (black) coverage is unaffected whereas with Kapa HiFi (red) and Collibri (green) coverage drops to near zero in the GC-rich region in the region in the centre of the plot. Likewise in (b) RepliQa (blue), Watchmaker (black) and PCR-free data (pink) coverage is unaffected whereas with Kapa HiFi (red) and Collibri (green) coverage drops to near zero in the AT-rich region in the region in the centre of the plot.

Fig. 5.. Low coverage index (fraction of genome covered at <50 % mean coverage) for amplification of 1 ng P falciparum Illumina library template with a range of enzymes using PCR conditions described by Lopez-Barragan [14] (blue), Quail (orange) [4], Oyola [12] (grey) and Aird [15] (yellow).