Generate phylogenetic tree from primary and metastatic tumors

Question

[Aidan.Matzko]

Hello,

I've obtained 8 samples, 1 from a primary tumor and 7 from metastatic tumors, and ran them through our DNAseq pipeline. Now, I want to visualize how the tumors evolved and create a phylogenetic tree that shows which tumors are closely related. I looked at previous posts, but they all seemed to be looking at subclones in a single tumor, where I'm looking at multiple tumors. Are there any tools I can use for this purpose?

Thanks so much!

(For reference, similar to figure 3C of this paper but using multiple tumors instead of one. http://www.nejm.org/doi/full/10.1056/NEJMoa1616288#t=article)

Answer 1

[poisonAlien]

There are quite a few programs available. Lichee does better job. It also incorporates allele frequencies rather than using just a binary matrix.

[Aidan.Matzko]

Lichee will work even with multiple tumors as opposed to multiple samples of one tumor?

[poisonAlien]

your 8 samples, are they from same patient ? i.e, multi region sequencing ?

[Aidan.Matzko]

Yes. They are from different tumors in the same patient (e.x one sample from the liver, one from the lung, one from the brain, etc).

Are there any tools that will generate more of a 'map'? Lichee was a nice start but wasn't exactly what I wanted.

Answer 2

[poisonAlien]

Okay I will post this as an answer.

I am not sure about any other tool. Another one I am aware also uses VAF for clustering and clonal composition. But I know what you mean, just like a phylogenetic tree.

This can be achieved with discrete character parcimony implmented in Phylip package. This method is as described in this and similar articles by Swanton's group.

You will have to create a binary table of all unique mutations detected across all tumor sections from a patient. e.g, If you have 5 samples from a patient (1 germline and 4 metastatic) and you detect 57 unique mutations from all 4 sections, table would look something like below:

5 57
T3  11111111111111111111111111111111111001111111111111111001111111111
T1  11111111111111111111111111111110110111111111111111111111111111110
T4  11111111111111111111111111111111111111111111100111110001111110101
T2  11111111111111111011111011111111111111111111111110111111111101011
N   00000000000000000000000000000000000000000000000000000000000000000

where N is your germline, which would be zero for all somatic mutations and acts as an outgroup root. T1 to T4 are your metastatic tumor sections (liver, brain , lung etc). 1 is presence and 0 is an absence of a mutation.

From terminal

phylip pars

and type your binary table name. Hit O and set 5 as an out group root (here is 5th is your germline). This should generate two output files called outtree and outfile. outfile is a newick format tree file. You can pass this to phylip drawtree or read into R using ape::read.tree and plot it with plot command. There is also ggtree to draw trees with ggplot.

Output from above exaple using phylip drawtree

from R ape: