Skip to main content
Figure 1 | Genome Biology

Figure 1

From: Mobster: accurate detection of mobile element insertions in next generation sequencing data

Figure 1

Overview of the Mobster algorithm. (A) In the first phase discordant ends (long red arrows) and clipped ends (short red arrows) are extracted from the BAM file when, respectively, the mate or the unclipped end is mapped uniquely to the reference. Subsequently these reads are mapped to the mobilome and investigated for having a polyA/T tail. (B) After mapping, all mates and unclipped sequences (anchors) belonging to unambiguously mapped Alu, L1, SVA, or HERV-K reads are identified. Anchors of clipped reads are clustered separately from anchors of discordant reads. (1) For clipped clusters, anchors should be: (i) supportive of the same ME family or same polyA/T stretch; (ii) clipped on the same side; and (iii) clipped within a few bp of each other. (2) The 5' clipped cluster (arrow pointing to right), consisting of right-clipped reads, and 3' clipped cluster (arrow pointing to left), consisting of left-clipped reads, are indicative of the same MEI event when: (i) they support the same ME family or one of the clusters supports a ME family and the other cluster supports a polyA/T tail; and (ii) they overlap by a maximum of 50 bp (allowing for TSDs) or are separated by a maximum of 20 bp (allowing for target site deletions). (3, 4) Discordant pair anchors, are clustered when: (i) they map to the same strand; (ii) are supporting the same ME family; and (iii) have start positions, which are within a specified neighborhood distance (4). (5) Forward strand anchors form 5' discordant clusters; reverse strand anchors form 3' discordant clusters. Discordant clusters from the 5' and 3' are indicative of the same MEI event when they overlap by maximal 50 bp or are within a user-defined window size. When possible, clipped clusters are merged with discordant clusters.

Back to article page