Analysis of intrapatient heterogeneity uncovers the microevolution of Middle East respiratory syndrome coronavirus

Park, Donghyun; Huh, Hee Jae; Kim, Yeon Jeong; Son, Dae-Soon; Jeon, Hyo-Jeong; Im, Eu-Hyun; Kim, Jong-Won; Lee, Nam Yong; Kang, Eun-Suk; Kang, Cheol In; Chung, Doo Ryeon; Ahn, Jin-Hyun; Peck, Kyong Ran; Choi, Sun Shim; Kim, Yae-Jean; Ki, Chang-Seok; Park, Woong-Yang

Analyzing the MERS-CoV genome sequences, we noticed that a significant number of nucleotide sites had mixed bases. Because deep sequencing not only generates a consensus sequence but also identifies non-consensus nucleotides at each position, Cotton et al. (2013a) analyzed nucleotide variants present at >1% frequency to determine the variants important for intrahost evolution. In the previous study, the nucleotide variants present at a frequency greater than the sequencing error rate (i.e., 1%) were considered to be true variants, although this estimate may be conservative for most applications. However, errors introduced during reverse transcription and PCR amplification can significantly distort estimates of allele frequencies, especially if a limited amount of RNA is used as an input. Thus, in this study, we evaluated the reproducibility of allele frequencies of non-consensus nucleotides. For this purpose, duplicate libraries generated by independent complementary DNA (cDNA) synthesis using the same RNA samples were sequenced and compared. Allele frequencies of duplicate samples from Patients 80 and 162 were significantly correlated, implying that the majority of mixed bases were a reflection of intrapatient heterogeneity rather than a result of sequencing errors (Fig. 1A ,B).