Shing Hei Zhan
Shing Hei Zhan
For input to BEAGLE, we have two gzipped VCF files, which we don't need to keep here, I think.
On a second thought, we might want to keep those VCF files, just to have a record of the input files to BEAGLE used to generate the forward and backward...
For the fwd and bwd matrices, BEAGLE has scaled and shifted values, e.g., `bwdVal[h] = scale*bwdVal[h] + shift`, where `h` is index of a haplotype in the ref. panel. I...
> I'd like the unit tests we do for unit tests to be even smaller - 3 or 4 samples and < 10 sites. How about this? 5 haploid individuals...
When we get the fwd and bwd matrices, we want one run with all the sites and then another run with chip-like sites, right? If so, probably we want like...
Ah, actually, scrap the above comment. BEAGLE just computes the values for the sites in the target VCF file. The values for the in-between sites are interpolated thereafer.
I've run BEAGLE 4.1 using a ref. panel of 2 diploid individuals (BEAGLE doesn't take haploid individuals, it seems) and a query set consisting of 1 diploid individual. This toy...
I'm storing everything in text and then convert to a Pandas dataframe or tree sequence.
Should we add an example with a simple genetic map? By default, BEAGLE 4.1 assumes that 1 Mb = 1 cM.
We should probably also add the state probability matrix that combines the forward and backward probabilities.