Cancer genomics and NGS

Question

[mst]

Hello,

I'm reading about cancer genomics and for better understanding of the subject decided to perform several numerical experiments. There are several naive questions that I would like to ask:

1) De novo assembly vs reads alignment - what is the preferred approach to assemble the whole tumor/normal genomes?

2) Is there any 'reads' simulator that allows to simulate cancer's genomes (with different types of aberrations, not only SNV) base on a reference one?

Thanks a lot, Maxim

Answer 1

[Chris Miller]

1) Alignment and assembly are two very different concepts, so this isn't really an answerable question.

2) https://github.com/adamewing/bamsurgeon

[mst]

Chris, thanks a lot for the link. By "reads alignment" I meant re-sequencing. So I thought to take a tool like bowtie, human reference genome and re-sequence "normal" genome of a donor. Then using this genome as a reference, re-sequence tumor genome of the same donor. Is it a correct route?

[Chris Miller]

You would benefit from reading some review articles on cancer genomics to get your terminology straight. Yes, the basic principle is to sequence both tumor and normal genomes from an individual, align them to the human reference, and compare the results to find somatic mutations

[jotan]

With the caveat that cancer genomes can be rearranged and have large CNVs.

Maybe this is why there is confusion regarding de novo assembly?

Some studies assemble cancer genomes de novo and map to that, rather than the normal reference.

[mst]

I'm still thinking about de novo/re-sequencing approaches. As I've read there could be different types of structural variations in cancer genomes, including inversions. Would it be useless to try assembly such genomes using re-sequencing?