Implement Cecile Ane's polyploidy method as a new CAFE feature

[benfulton]

It would be interesting to implement the method Ane developed in Rabier et al. (2014). Ane has given permission and we need to make sure to give credit where due.

https://www.ncbi.nlm.nih.gov/pubmed/24361993

Software: http://www.stat.wisc.edu/~ane/wgd/

[benfulton]

Installed R 3.4 to /usr/local/lib/R/bin, then library("phyext") library("WGDgc") tre.string = "(D:{0,18.03},(C:{0,12.06},(B:{0,7.06},A:{0,7.06}):{0,2.49:wgd,0:0,2.50}):{0, 5.97});" tre.phylo4d = read.simmap(text=tre.string) dat = data.frame(A=c(2,2,3,1), B=c(3,0,2,1), C=c(1,0,2,2), D=c(2,1,1,1)); a = processInput(tre.phylo4d, startingQ=0.9) getLikGeneCount(log(c(.01,.02)),a,dat,mMax=8,geomProb=1/1.5,conditioning="oneOrMore")

[josiahseaman]

With permission, I would like to implement this feature as part of my PhD Thesis. I need the gene count history information output by Cafe plus the WGD handling of Ane and Rabier. Also, I think it's good to consolidate features in well-supported bioinformatics tools. Cafe is already a great program, and I'd like to contribute. I have 3 years of experience in C++ and 15 years of experience in other languages. I plan on forking the repo and submitting a pull request when I'm done.

Key points to verify before I get started

1. How much overlap is there in the underlying mathematics of Cafe and WGDgc is there?
2. Is it sufficient to model a WGD as an instantaneous spike in the birth parameter λ, while µ = 0?
3. Will Cafe handle a tree containing an internal line singleton of size 0? This is the way it's annotated in the WGDgc SIMMAP format.
   • Possible solution is a length = 1 line segment. But this isn't great for polyploids <1Mya.
4. Will λ =1 for length 1 mean every gene family is doubled, or is it probabilistic growth with replacement allowing for 3x and 4x duplicates?

Would it be acceptable for me to implement this feature?

[benfulton]

Hi,

We would love to have you work on this. We are planning on a new codebase for CAFE that would be more appropriate for this work. I suggest we work together on a branch in the new repository as I suspect it will be a multiple-commit process. Send me a note (befulton at iu dot edu).