Author:
Drmanac Snezana,Callow Matthew,Chen Linsu,Zhou Ping,Eckhardt Leon,Xu Chongjun,Gong Meihua,Gablenz Scott,Rajagopal Jyothi,Yang Qing,Villarosa Christian,Au Anthony,Davis Kyle,Jorjorian Alexander,Wang Jingjing,Chen Ao,Zhang Xian,Borcherding Adam,Wei Xiaofang,Zhang Mingxuan,Xie Yonghui,Barua Nina,Shafto Jay,Dong Yuliang,Zheng Yue,Wang Lin,Zhai Lili,Li Jiguang,Liao Sha,Zhang Wenwei,Liu Jian,Jiang Hui,Wang Jian,Li Handong,Xu Xun,Drmanac Radoje
Abstract
AbstractMassively parallel sequencing (MPS) on DNA nanoarrays provides billions of reads at relatively low cost and enables a multitude of genomic applications. Further improvement in read length, sequence quality and cost reduction will enable more affordable and accurate comprehensive health monitoring tests. Currently the most efficient MPS uses dye-labeled reversibly terminated nucleotides (RTs) that are expensive to make and challenging to incorporate. Furthermore, a part of the dye-linker (scar) remains on the nucleobase after cleavage and interferes with subsequent sequencing cycles. We describe here the development of a novel MPS chemistry (CoolMPS™) utilizing unlabeled RTs and four natural nucleobase-specific fluorescently labeled antibodies with fast (30 sec) binding. We implemented CoolMPS™ on MGI’s PCR-free DNBSEQ MPS platform using arrays of 200nm DNA nanoballs (DNBs) generated by rolling circle replication and demonstrate 3-fold improvement in signal intensity and elimination of scar interference. Single-end 100-400 base and pair-end 2×150 base reads with high quality were readily generated with low out-of-phase incorporation. Furthermore, DNBs with less than 50 template copies were successfully sequenced by strong-signal CoolMPS™ with 3-times higher accuracy than in standard MPS. CoolMPS™ chemistry based on natural nucleobases has potential to provide longer, more accurate and less expensive MPS reads, including highly accurate “4-color sequencing” on the most efficient dye-crosstalk-free 2-color imagers with an estimated sequencing error rate of 0.00058% (one error in 170,000 base calls) in a proof-of-concept demonstration.
Publisher
Cold Spring Harbor Laboratory