Stdpopsim: Kamm QC

[apragsdale]

QC for Kamm et al. (2009) model, see PR #146. Also see discussion regarding ancient samples in Issue #166, which should be addressed before closing this QC.

[apragsdale]

I added QC code for this model in PR #169. There were a few things that I was confused about that might lead to differences between the model implementations:

1. I couldn't find a published generation time - I think this is because they rescaled time to years based on the known dates of the ancient samples, so that all inferred dates were relative to those known dates.
2. The size history function for the Neanderthal branch is not well described. The Neanderthal pop has constant size equal to the ancestral size until the time of the Mbuti and Han branches merge, at which point it declines exponentially to the inferred size N_Nean - but it isn't clear at what time that branch has that size. Is it at the time of sampling the Altai individual? That would make the most sense to me, and it's what I went with.

[ndukler]

Who implemented the original model?

[apragsdale]

@FerRacimo did a few weeks ago. I didn't look through his implementation until this morning, but looks like there are differences in how size history along the Neanderthal branch was set up, and choice in generation time.

[FerRacimo]

Yep, it was not clear to me how they were setting N_Nean. There were also some other arbitrary choices I had to make (see comments inside the history) because the parameter values were not specified in the paper. Also, I had to set the dates of all ancient populations to the present in the current version of the model (which is obviously not how it's done in the paper) because of Issue #166.

[apragsdale]

Exactly what I was running into. For the most part, it looks like we interpreted the model the same way. I'll give the paper one more close read-through to see if I missed the description of the neanderthal branch buried in the appendix or something. I think there was an exponential decay in pop size, I'm just not sure when the decline ended (what time it had size N_Nean).

[ndukler]

It looks like it might be worth asking the first author for some clarification.

[apragsdale]

Ok - so I emailed with Jack about this to figure out the size function along the neanderthal branch. N_Nean is the size at the time of sampling the Altai individual (45ka). For generation time, I think we should just use the default generation time, since they did not need to specify one to rescale from genetic units to years.

We'll need to make sure the model and QC version match, and also handles sampling ancient populations however it get resolves in #177. For now I'll go back through the QC model I wrote up to double check it follows the Neanderthal pop size and default generation time.

[FerRacimo]

Did he specify what the size was before the bottleneck and if it follows
some sort of exponential decay?

By default gen time, I'm guessing that's 29?

On Wed, Nov 20, 2019 at 9:22 PM Aaron Ragsdale [e-mail]
wrote:

Ok - so I emailed with Jack about this to figure out the size function
along the neanderthal branch. N_Nean is the size at the time of sampling
the Altai individual (45ka). For generation time, I think we should just
use the default generation time, since they did not need to specify one to
rescale from genetic units to years.

We'll need to make sure the model and QC version match, and also handles
sampling ancient populations however it get resolves in #177
https://github.com/popgensims/stdpopsim/pull/177. For now I'll go back
through the QC model I wrote up to double check it follows the Neanderthal
pop size and default generation time.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/popgensims/stdpopsim/issues/167?email_source=notifications&email_token=ABLXSAXUGGJKSYJJLS53POTQUWL75A5CNFSM4JL2U3EKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEEUSQSI#issuecomment-556345417,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/ABLXSASNTGQGRF2ABJCPHH3QUWL75ANCNFSM4JL2U3EA
.

--
Fernando Racimo, PhD
Assistant Professor
Centre for GeoGenetics
University of Copenhagen
Øster Voldgade 5-7, DK-1350 Copenhagen, Denmark

[apragsdale]

Yeah, the size before the decay began is the the same as the ancestral human size (N_Nean_Losch), with decay starting at the same time that Mbuti and Eurasians split.

Right now it looks like the human default generation time is set to 25 years in homo_sapiens.py.

[FerRacimo]

Ok cool, I'll modify the original model then.

On Wed, Nov 20, 2019 at 9:43 PM Aaron Ragsdale [e-mail]
wrote:

Yeah, the size before the decay began is the the same as the ancestral
human size (N_Nean_Losch), with decay starting at the same time that Mbuti
and Eurasians split.

Right now it looks like the human default generation time is set to 25
years in homo_sapiens.py.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/popgensims/stdpopsim/issues/167?email_source=notifications&email_token=ABLXSAQP3U34YTOF6WXYJJLQUWOO3A5CNFSM4JL2U3EKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEEUYM3I#issuecomment-556369517,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/ABLXSAQN4W5HXRCNNZZQCR3QUWOO3ANCNFSM4JL2U3EA
.

--
Fernando Racimo, PhD
Assistant Professor
Centre for GeoGenetics
University of Copenhagen
Øster Voldgade 5-7, DK-1350 Copenhagen, Denmark