search

Illumina / paragraph

Graph realignment tools for structural variants
Other
150 stars 28 forks source link

SV calling and genotyping across a population #50

Closed biozzq closed 4 years ago

biozzq commented 4 years ago

Hi @traxexx

I wonder how do you generate the candidate SVs that used as input for paragraph when you have many short-read samples and several representative long-read samples. In my mind, I will call SVs using both long-read and short-read samples and then merge them together. During merging, I think the breakpoints will become coarse. Does this affect the genotype results when I do not have a precise breakpoint?

Sincerely, Zheng Zhuqing

traxexx commented 4 years ago

As mentioned in #52 merging calls across different samples could be a problem. That's also the effort that people made in SURVIVOR. To answer your question: does this affect genotyping result? The answer is yes. Paragraph is a breakpoint genotype. Inaccurate breakpoints will result in a worse genotyping performance. We roughly quantified this effect in our paper, too.

biozzq commented 4 years ago

Thanks, I noticed the figure 3 in your paper.