Improved data analysis for the MinION nanopore sequencer.
Improved data analysis for the MinION nanopore sequencer.
复制标题
DOI：
10.1038/nmeth.3290
复制
发表时间：
2015-04
期刊：
影响因子：
48
通讯作者：
Akeson M
中科院分区：
文献类型：
--
作者：
Jain M;Fiddes IT;Miga KH;Olsen HE;Paten B;Akeson M
牛津纳米孔小人使用一系列孔对单个DNA分子进行测序，这些孔根据离子电流步骤读取核苷酸同一性。我们使用M13基因组dsDNA评估和优化了Minion的性能。使用期望最大化(EM)方法，我们得到了读出插入、删除和替换错误率的稳健最大似然估计(分别为4.9%、7.8%和5.1%)。我们发现，99%的高质量的2D附属物阅读映射到参考文献的平均同一性为85%。我们提出了一个针对单核苷酸变异(SNV)检测的小程序定制工具，该工具使用ML参数估计和对许多可能的读取比对进行边际化，以实现高达99%的精度和召回率。通过将我们的高置信度比对策略与长序列读取配对，我们解决了人类染色体Xq24未解析区域内的癌症/睾丸基因家族(CT47)的拷贝数。
The Oxford Nanopore MinION sequences individual DNA molecules using an array of pores that read nucleotide identities based on ionic current steps. We evaluated and optimized MinION performance using M13 genomic dsDNA. Using expectation-maximization (EM) we obtained robust maximum likelihood (ML) estimates for read insertion, deletion and substitution error rates (4.9%, 7.8%, and 5.1% respectively). We found that 99% of high-quality ‘2D’ MinION reads mapped to reference at a mean identity of 85%. We present a MinION-tailored tool for single nucleotide variant (SNV) detection that uses ML parameter estimates and marginalization over many possible read alignments to achieve precision and recall of up to 99%. By pairing our high-confidence alignment strategy with long MinION reads, we resolved the copy number for a cancer/testis gene family (CT47) within an unresolved region of human chromosome Xq24.