State Key Lab of Bioelectronics, School of Biological Science and Medical Engineering, Southeast University, Nanjing 210096, China.
Int J Mol Sci. 2019 Apr 21;20(8):1953. doi: 10.3390/ijms20081953.
As the most widely-used single cell whole genome amplification (WGA) approach, multiple displacement amplification (MDA) has a superior performance, due to the high-fidelity and processivity of phi29 DNA polymerase. However, chimeric reads, generated in MDA, cause severe disruption in many single-cell studies. Herein, we constructed ChimeraMiner, an improved chimeric read detection pipeline for analyzing the sequencing data of MDA and classified the chimeric sequences. Two datasets (MDA1 and MDA2) were used for evaluating and comparing the efficiency of ChimeraMiner and previous pipeline. Under the same hardware condition, ChimeraMiner spent only 43.4% (43.8% for MDA1 and 43.0% for MDA2) processing time. Respectively, 24.4 million (6.31%) read pairs out of 773 million reads, and 17.5 million (6.62%) read pairs out of 528 million reads were accurately classified as chimeras by ChimeraMiner. In addition to finding 83.60% (17,639,371) chimeras, which were detected by previous pipelines, ChimeraMiner screened 6,736,168 novel chimeras, most of which were missed by the previous pipeline. Applying in single-cell datasets, all three types of chimera were discovered in each dataset, which introduced plenty of false positives in structural variation (SV) detection. The identification and filtration of chimeras by ChimeraMiner removed most of the false positive SVs (83.8%). ChimeraMiner revealed improved efficiency in discovering chimeric reads, and is promising to be widely used in single-cell sequencing.
作为最广泛使用的单细胞全基因组扩增（WGA）方法，多重置换扩增（MDA）具有优越的性能，这要归功于 phi29 DNA 聚合酶的高保真度和高进程性。然而，MDA 产生的嵌合reads 会严重干扰许多单细胞研究。在此，我们构建了 ChimeraMiner，这是一种用于分析 MDA 测序数据的改进的嵌合reads 检测管道，并对嵌合序列进行了分类。使用了两个数据集（MDA1 和 MDA2）来评估和比较 ChimeraMiner 和以前的管道的效率。在相同的硬件条件下，ChimeraMiner 仅花费 43.4%（MDA1 为 43.8%，MDA2 为 43.0%）的处理时间。分别有 2440 万（6.31%）对读段和 1750 万（6.62%）对读段被 ChimeraMiner 准确分类为嵌合体。除了发现以前的管道检测到的 83.60%（17639371 个）嵌合体之外，ChimeraMiner 还筛选出 6736168 个新的嵌合体，其中大部分被以前的管道所忽略。在单细胞数据集上应用时，每个数据集都发现了所有三种类型的嵌合体，这在结构变异（SV）检测中引入了大量的假阳性。ChimeraMiner 通过对嵌合体的识别和过滤，去除了大部分假阳性的 SV（83.8%）。ChimeraMiner 在发现嵌合reads 方面显示出了更高的效率，有望在单细胞测序中得到广泛应用。
