Review lphy datatype naming, error message: Unphased genotype not supported for error model.

[EvaLiyt]

When we pass in unphased genotype for GT16 error model, it would give us the error that Error model cannot handle data type nucleotideDiploid10 while running lphybeast.

Here's the lphy script:

data {
   // alignment
    A = readFasta(file = "alignment_r0.fasta", sequenceType=unphasedGenotype());
    L = A.nchar();
}
model{
    lambda ~ LogNormal(meanlog = -0.6, sdlog = 0.1);
    mu ~ LogNormal(meanlog = -5.8, sdlog=0.1);
    tree ~ Yule(lambda=lambda, n=40);

    // gt16 model
    π ~ Dirichlet(conc=[3.0,3.0,3.0,3.0,3.0,3.0,3.0,3.0,3.0,3.0,3.0,3.0,3.0,3.0,3.0,3.0]);
    rates ~ Dirichlet(conc=[1.0, 2.0, 1.0, 1.0, 2.0, 1.0]);
    Q = gt16(freq=π, rates=rates);

    D ~ PhyloCTMC(L=L, Q=Q, tree=tree, dataType=phasedGenotype(), mu=mu);
    U = unphase(D);

    epsilon ~ Beta(alpha=2, beta=18);
    delta ~ Beta(alpha=1.5, beta=4.5);
    A ~ GT16ErrorModel(delta=delta, epsilon=epsilon, alignment=U);
}

Here is the fasta file: (change the extension before use it)

alignment_r0.txt

[kche309]

Hi @EvaLiyt, thanks for finding this bug. We'll aim to have this fixed as soon as we can.

The beast error model can work with unphased data if it is encoded as nucleotideDiploid16. I suspect this issue is because of the recent changes in the datatype mapping unphased genotypes to nucleotideDiploid10 by default.

@yaox1818 we should discuss the design of this before we attempt the fix. We also need to add some unit tests for lphybeast and the two diploid datatypes nucleotideDiploid10 and nucleotideDiploid16.

[kche309]

Hi Eva, can you try readFasta() with sequenceType=PhasedGenotype?

[kche309]

Hi @EvaLiyt and @yaox1818 we should review the naming of the lphy datatype classes to avoid confusion.
https://github.com/bioDS/beast-phylonco/tree/master/phylonco-lphy/src/main/java/phylonco/lphy/evolution/datatype

Current classes:

PhasedGenotype.java 
PhasedGenotypeFunction.java 
PhasedGenotypeState.java 
UnphasedGenotype.java 
UnphasedGenotypeFunction.java 
UnphasedGenotypeState.java

The PhasedGenotype encodes the diploid genotypes as 16 states, but the data could potentially be phased or unphased. I suggest we rename it to DiploidGenotype16 or Genotype16. The BEAST equivalent is NucleotideDiploid16 - any comments or suggestions?

For UnphasedGenotype - perhaps DiploidGenotype10 or Genotype10? The BEAST equivalent is NucleotideDiploid10.

If we go forward with the changes, we also need to deprecate the usage of datatype=PhasedGenotype() and UnphasedGenotype() in lphy scripts, and update scripts that use these functions.