yangjl / dadi Goto Github PK

In the examples/fs_from_data directory, I include an example of loading a 
spectrum from a raw 
SNP data file. I should standardize this file format, and include the relevant 
code in the main line of 
dadi.

Working with a folded Spectrum is at present a little tricky. In particular, 
refolding a folded 
Spectrum wipes out almost all entries, and this can be done implicitly in the 
optimizer.

I think the proper solution is to have Spectrum objects record whether or not 
they are folded.

1. Add an isfolded member variable to Spectrum objects. (Use Python property 
functionality?)
2. Record isfolded status in the saved files.
3. Make optimizer handle folding automatically, folding the model result if the 
data is folded, for 
example.
4. It doesn't make sense to project a folded Spectrum, because during 
projection it is possible for 
minor allele status to switch. To deal with this, we need an unfold() method 
that will assume each 
allele is minor with 50% probability. Then we can raise an exception when 
projecting a folded 
Spectrum.

The website would benefit from a running list of papers dadi has been used for.

This is an issue in bootstrapping, as raised by Jeffrey Lozier.

In plant populations, self-fertilization can be the dominant reproductive 
process. Does selfing 
change results from a diffusion-based analysis?

I have code for measuring the decay of LD (r^2) in ms simulations. This is 
useful to have for 
validating the model, so I should include it in the dadi distribution.