I am designing an experiment where I will need to construct a tumor phylogenetic tree from different areas of a primary tumor.
There are two critical hurdles you're going to face when doing this analysis:
1) The vast majority of your somatic mutations are going to be common to all samples. This means that you're not going to have very many differentiating mutations for each tumour region.
2) Your tumour samples are each going to admixtures of different subclones Your phylogenetic analysis needs to account for the heterogeneity within each sample.
from different areas of a primary tumor
The somatic mutations in primary and mets are almost the same (https://www.sciencedirect.com/science/article/pii/S2211124719305789#fig2). Within a primary you're going to have hardly any differences.
I am leaning towards the WGS
I strongly recommend WGS for this style of analysis and including SVs and CNVs in your variant analysis. SVs/CNVs give important information about tumour phylogeny and, when you do joint variant calling with callers such as GRIDSS, you can detect low subclonality (<5%) SVs with more sensitivity than you can SNVs.
Should I use WGS (60x) or a deeper WES ( > 100x) to sequence these areas
If you want to justify deep WGS sequencing, then I recommend finding some papers that quantify the expected number of somatic mutations that are actually different. Even if you just use primary vs mets (I found that paper in a couple of minutes of googling), at least you'll have a upper bound of how well you'll be able to construct your tree. If you think the difference within primary is going to be much less than primary vs met (which it should be), then you'll need to be much more sensitive. I expect you'll definitely need WGS, and maybe even deeply sequenced WGS but you should run the numbers to get a vague idea of how many mutations you're actually expecting.
