Optimal enzymes for amplifying sequencing libraries

Abstract

PCR amplification introduces bias into Illumina sequencing libraries1. Although amplification-free library preparation solves this, micrograms of starting material are usually required. Most researchers follow standard protocols using Phusion polymerase, which has processivity and fidelity advantages over most polymerases. Yet for genomics applications, our demands on DNA amplification systems often surpass their specification. Thermostable DNA polymerases such as Phusion are used to amplify mixtures of fragments, albeit with variable efficiency. Typically, (G+C)-neutral fragments are amplified with higher efficiency than extremely (G+C)-rich or (A+T)-rich fragments. The accumulation of these slight differences in amplification over multiple cycles often results in profound bias. There have been reports of using alternative DNA polymerases for Illumina library construction2, 3, 4, but these are infrequent, and comprehensive analyses are lacking. To reduce bias, we investigated many thermostable DNA polymerases and alternate reaction conditions for amplification of adapter-ligated fragments for Illumina sequencing. We expect this comparison to be relevant to other applications that involve simultaneous amplification of complex fragment mixtures.