From 0cd2752b27526c4417065b14e99b5ebe793321be Mon Sep 17 00:00:00 2001
From: daikitag <48062118+daikitag@users.noreply.github.com>
Date: Fri, 16 Feb 2024 15:13:32 +0000
Subject: [PATCH] DOC: Freq-dep

Modify the frequency dependence explanation in the tstrait documentation, as it is not reflecting the truth.
---
 docs/effect-size.md      | 11 ++---------
 docs/genetic.md          |  1 -
 tstrait/genetic_value.py |  2 --
 3 files changed, 2 insertions(+), 12 deletions(-)

diff --git a/docs/effect-size.md b/docs/effect-size.md
index 909f4f4..b21cdf8 100644
--- a/docs/effect-size.md
+++ b/docs/effect-size.md
@@ -207,7 +207,7 @@ The details of sites in tree sequences can be found [here](tskit:sec_site_table_
 
 (trait_frequency_dependence)=
 
-## Frequency dependence
+## Frequency Dependence
 
 Tstrait supports frequency dependence simulation. It has been shown that rare variants
 have increased effect sizes compared with common variants
@@ -220,7 +220,7 @@ In the frequency dependence model, the following value is multiplied to the effe
 
 ```{math}
 :label: eq:freq-dep
-[2p(1-p)^\alpha]
+\Big[\sqrt{2p(1-p)}\Big]^\alpha
 ```
 
 In the above expression, $p$ is the frequency of the causal allele, and
@@ -235,13 +235,6 @@ ignore the `alpha` parameter if you are not interested in implementing the
 frequency dependent model.
 :::
 
-The frequency dependence architecture is still an ongoing research topic. While the
-frequency dependence model can be used for any trait models in tstrait, it is
-suggested that you use the trait model with mean 0 and `alpha` to be -1/2 to conduct
-simulations that are widely used in
-simulation-based research projects (See
-[Speed et al. (2017)](https://doi.org/10.1038/ng.3865) for details).
-
 In the below example, we will be demonstrating how `alpha` influences the simulated
 effect sizes by using a simulated tree sequence with 10,000 individuals.
 
diff --git a/docs/genetic.md b/docs/genetic.md
index a78043f..c510d20 100644
--- a/docs/genetic.md
+++ b/docs/genetic.md
@@ -29,7 +29,6 @@ After this genetic value page, you will be able to:
 
 - Understand how to generate genetic value in tstrait
 - Understand how to use the user's defined effect sizes to generate genetic values
-- Understand the details of frequency dependence model that is supported in tstrait
 
 # Algorithm Overview
 
diff --git a/tstrait/genetic_value.py b/tstrait/genetic_value.py
index bc90e09..3d8ea57 100644
--- a/tstrait/genetic_value.py
+++ b/tstrait/genetic_value.py
@@ -51,8 +51,6 @@ class _GeneticValue:
     trait_df : pandas.DataFrame
         Dataframe that includes causal site ID, causal allele, simulated effect
         size, and trait ID.
-    alpha : float
-        Parameter that determines the relative weight on rarer variants.
     """
 
     def __init__(self, ts, trait_df):