Msprime multiple roots?

[NewAtl]

Hi everyone!

I am working on a project that involves simulating methylation states based on the recombination patterns generated by msprime. My objective is to analyze the output to understand how recombination affects methylation across the genome. I'm still very new to the program, so I would appreciate some guidance :D

Problem

I am trying to read the nodes and edges tables from msprime to analyze the tree structures they represent. However, I am encountering an issue:

• Issue: Despite setting various parameters, I consistently end up with multiple trees in the output tables. This means that:
  • I identify the root of the tree at the earliest time point present in the nodes table.
  • However, a new root appears at a later time point, which is not connected to the initial tree formed under the earliest root.

Question

I believe I am misunderstanding some aspect of how msprime handles tree sequences or how I should interpret them. Specifically, I am confused about:

• Why do multiple roots (trees) appear in the output?
• How can I correctly interpret or set parameters to avoid this situation?

Thank you for your help! If you need more information about the parameters I am using or specific examples from the output, please let me know.

This is one of my tries to get the needed tables:
ts = msprime.sim_ancestry( 10, recombination_rate=0.5, sequence_length=20,additional_nodes=msprime.NodeType.RECOMBINANT,coalescing_segments_only=False )

[molpopgen]

Due to recombination, the root can change along the genome. In other words, different segments of the genome can have different MRCA nodes. But with msprime, each tree will have one root:

>>> ts = msprime.sim_ancestry( 10, recombination_rate=0.5, sequence_length=20,additional_nodes=msprime.NodeType.RECOMBINANT,coalescing_segments_only=False )
>>> for t in ts.trees():
...     print(t.num_roots)
... 
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1

[molpopgen]

For example, see here for some inspiration.

[NewAtl]

Ah thank you! Do i understand that correct, that witht the argument record_full_arg=True the tables from msprime are representing an ancestral recombination graph? wouldnt that be the graph with only one root?

[molpopgen]

yes -- at this point, it may be most useful to recreate the examples from that tutorial in, for example, a jupyter notebook, to help visualize the relationship b/w the options and the output.

[jeromekelleher]

wouldnt that be the graph with only one root?

The graph doesn't have to have one root though (in fact won't most of the time), because you will have different MRCA nodes at different points along the sequence. To have one root you would need to simulate back to the "Grand MRCA", which takes exponential time.

See the "Big ARG and little ARG" section in our recent Genetics paper this point, and for general background on the difference between the ARG as a graph, and a succinct tree sequence. It's quite subtle, so I wouldn't expect to understand immediately.

[molpopgen]

Thanks @jeromekelleher. @NewAtl -- I didn't look closely enough at the examples in the ARG tutorial and I missed that there is a mouse-over action that shows you each embedded marginal tree.