BUG: Effect size simulation

[daikitag]

The original code was dividing the effect sizes by the number of causal sites, which led to the simulation of traits having extremely small variance.

I modified the codes for all trait models in tstrait, such that the mean and variance of the overall phenotype are balanced.

I would really like to thank Alison for spotting this by reading the tstrait manual.

[jeromekelleher]

I think we need to do a few more things @daikitag as this is a serious bug:

1. Create an issue to track the bug that we can refer to (not this PR). This should describe the bug clearly and precisely, and document any effects on downstream users
2. Update the manual
3. Update the changelog to briefly describe the bug, and provide a link to corresponding issue and PR(s)

[daikitag]

I think we need to do a few more things @daikitag as this is a serious bug:

1. Create an issue to track the bug that we can refer to (not this PR). This should describe the bug clearly and precisely, and document any effects on downstream users
2. Update the manual
3. Update the changelog to briefly describe the bug, and provide a link to corresponding issue and PR(s)

Thanks Jerome. I will dramatically update the manual for effect size simulation in the next PR.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (5371a97) 100.00% compared to head (f175d07) 100.00%.

Additional details and impacted files

@@            Coverage Diff            @@
##              main      #100   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            8         8           
  Lines          302       302           
  Branches        31        31           
=========================================
  Hits           302       302           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[jeromekelleher]

LGTM, but I didn't realise this was a problem in the first place and it's not obvious to me that we've solved it.

I think the deeper problem here is that we are not comparing our results against other programs. We should be doing this - I'll open an issue to track.

[jeromekelleher]

Thanks Jerome. I will dramatically update the manual for effect size simulation in the next PR.

Can you open an issue to track that then please? We want a set of modular issues that we can address before we release the update .

[daikitag]

LGTM, but I didn't realise this was a problem in the first place and it's not obvious to me that we've solved it.

I think the deeper problem here is that we are not comparing our results against other programs. We should be doing this - I'll open an issue to track.

I agree. I will examine how other simulators are simulating effect sizes, and I will let you know after I figure it out.

[daikitag]

Removing this pull request, due to changes made in #107. This architecture will be implemented following the feedback in #101.