AccuNGS: a sequencing protocol for detection of ultra-rare variants reveals extensive variation in HIV during first days of infection

Next generation sequencing (NGS) is widely used to identify genetic diversity within a sample. This includes basic scientific interrogations such as quantifying standing genetic variation and estimating mutation rates, as well as clinical applications such as surveillance of drug-resistance mutations and assessment of a pre-tumorigenic state. While several novel sequencing methods were established for accurately measuring rare genetic variants (prevalence below 1:100), the low yield of these methods and their relative complexity limits their utility for clinical samples. Here we describe a simple NGS protocol with high template recovery that enables detection of variants at frequencies as rare as 1:10,000. We thoroughly explored possible contributors to the background error level of the protocol by performing differential sequencing that targeted virtually every step of the protocol. Our results reveal that unlike most protocols in use, errors introduced during sample amplification are not predominant. We highlight that as opposed to what is commonly assumed, Illumina MiSeq paired-end reads are not independent from each other, and biochemical processes inside the sequencer itself likely contribute most to the background errors. Finally, we applied our method to an RNA sample from a seronegative HIV-1 patient during acute infection, i.e., with an infection duration of approximately twenty days. Our results revealed transition mutations in around 50% of the positions sequenced with elevated rates of G>A mutations, possibly driven by host deaminases. These results highlight the early rich accumulation of genetic diversity in early HIV infection at depths previously unseen.