When calling and genotyping SVs among very closely related populations, the false positive rate is very high.

[kuangzhuoran]

I am using a workflow of manta + svimmer + graphtyper to identify structural variants (SVs) from short read whole genome resequencing data.

I have applied this pipeline to four types of datasets: (1) comparisons between different species (wild rodents), (2) domesticated cattle versus wild cattle, (3) comparisons between different species (wild birds), and (4) comparisons among different populations within the same species (wild birds, with an estimated divergence time of around 10,000 years).

The pipeline performs well on the first three types of datasets. However, for different populations within the same species (wild birds), it produces a large number of false positives, which is inconsistent with manual inspection in IGV.

[kuangzhuoran]

In the first three cases, the divergence between groups is substantial, and artificial domestication can act as an accelerator of evolution. This is not the case for the fourth dataset.