The fingerprint of phantom mutations in mitochondrial DNA data

Abstract

Phantom mutations are systematic artifacts generated in the course of the sequencing process itself. In sequenced mitochondrial DNA (mtDNA), they generate a hotspot pattern quite different from that of natural mutations in the cell. To identify the telltale patterns of a particular phantom mutation process, one first filters out the well-established frequent mutations (inferred from various data sets with additional coding region information). The filtered data are represented by their full (quasi-)median network, to visualize the character conflicts, which can be expressed numerically by the cube spectrum. Permutation tests are used to evaluate the overall phylogenetic content of the filtered data. Comparison with benchmark data sets helps to sort out suspicious data and to infer features and potential causes for the phantom mutation process. This approach, performed either in the lab or at the desk of a reviewer, will help to avoid errors that otherwise would go into print and could lead to erroneous evolutionary interpretations. The filtering procedure is illustrated with two mtDNA data sets that were severely affected by phantom mutations.

Figure 1

The full quasi-median networks representing the weighty variation within 16051–16365 (i.e., not scoring the speedy transitions listed in the “Weighty HVS-I Filter” subsection of the “Methods” section) for the following three subsets of the Brazilian mtDNA data (Alves-Silva et al. 2000): American (a), African (b), and European (c). Every unit-length link signifies one weighty transition, except where “t” indicates a transversion. The size of the nodes is proportional to the number of sampled mtDNAs of this haplotype; n is the sample size and h the number of sampled haplotypes (relative to the weighty mutations).

Figure 2

The full median networks representing the weighty variation in the Japanese mtDNA data (Seo et al. 1998) for the mtDNAs with 16189C (yielding a long C run) (a) and 16189T (b). Symbols are as in figure 1.

Figure 3

The full median network representing the weighty variation in the Ladin mtDNA data (Stenico et al. 1996). Symbols are as in figure 1.

Figure 4

Portion of the chromatogram of a mtDNA sequence exhibiting the A→T 16220 artifact.