phenotypes not initialized from TreeSeq file

[petrelharp]

Suppose we run this script:

initialize() {
	initializeTreeSeq();
	initializeMutationRate(1e-3);
	initializeMutationType("m1", 0.5, "f", 0.001);
	initializeGenomicElementType("g1", m1, 1.0);
	initializeGenomicElement(g1, 0, 99999);
	initializeRecombinationRate(1e-8);
}

1 early() {
	sim.addSubpop("p1", 10);
}

early() {
	cat("early: ");
	catn(p1.individuals.phenotypeForTrait());
}

1:19 late() {
	cat("late: ");
	catn(p1.individuals.phenotypeForTrait());
}

20 early() {
	cat("final: ");
	catn(p1.individuals.phenotypeForTrait());
	sim.treeSeqOutput("in.trees"); 
}

This writes out (in early()!) the tree sequence file with phenotypes recorded in metadata; we can verify this as follows:

# python -c "import tskit; print([ind.metadata['per_trait'] for ind in tskit.load('in.trees').individuals()])"
[[{'phenotype': 6.7861415138917165, 'offset': 1.0}], [{'phenotype': 6.67690035428362, 'offset': 1.0}], [{'phenotype': 6.59069003021617, 'offset': 1.0}], [{'phenotype': 6.602123797948265, 'offset': 1.0}], [{'phenotype': 6.725359768171507, 'offset': 1.0}], [{'phenotype': 6.691827682261045, 'offset': 1.0}], [{'phenotype': 7.033875325262007, 'offset': 1.0}], [{'phenotype': 7.005805053562283, 'offset': 1.0}], [{'phenotype': 6.67512438707715, 'offset': 1.0}], [{'phenotype': 6.51362239332308, 'offset': 1.0}]]

However, loading this back in with:

initialize() {
	initializeTreeSeq();
	initializeMutationRate(1e-3);
	initializeMutationType("m1", 0.5, "f", 0.001);
	initializeGenomicElementType("g1", m1, 1.0);
	initializeGenomicElement(g1, 0, 99999);
	initializeRecombinationRate(1e-8);
}

1 early() {
	sim.readFromPopulationFile("in.trees");
	cat("read in: ");
	catn(p1.individuals.phenotypeForTrait());
}

shows us that the individual's phenotypes are all NAN.

I think this is a bug?

[bhaller]

Sounds like a bug, yep! I'll look into it, thanks!