@@ -310,11 +310,11 @@ What are type I and type II risks?
...
@@ -310,11 +310,11 @@ What are type I and type II risks?
<details><summary>Solution</summary>
<details><summary>Solution</summary>
<p>
<p>
- Type I risk: $\alpha = \PP(\text{reject } \H_0\ \vert\ \H_0 \text{ is true}$
- Type I risk: $\alpha = \PP(\text{reject } \H_0\ \vert\ \H_0 \text{ is true})$
- Type II risk: $\beta = \PP(\text{not reject } \H_0\ \vert\ \H_0 \text{ is false}$
- Type II risk: $\beta = \PP(\text{not reject } \H_0\ \vert\ \H_0 \text{ is false})$
- Power $= 1 - \beta = \PP(\text{reject } \H_0\ \vert\ \H_0 \text{ is false}$
- Power $= 1 - \beta = \PP(\text{reject } \H_0\ \vert\ \H_0 \text{ is false})$
---
---
...
@@ -614,7 +614,7 @@ Imagine a procedure based on simulation to estimate the power of the previous T-
...
@@ -614,7 +614,7 @@ Imagine a procedure based on simulation to estimate the power of the previous T-
<details><summary>Solution</summary>
<details><summary>Solution</summary>
<p>
<p>
The power of a test is $1 - \beta = \PP(\text{reject } \H_0\ \vert\ \H_0 \text{ is false}$ where $\beta = \PP(\text{not reject } \H_0\ \vert\ \H_0 \text{ is false}$ is the type II risk.
The power of a test is $1 - \beta = \PP(\text{reject } \H_0\ \vert\ \H_0 \text{ is false})$ where $\beta = \PP(\text{not reject } \H_0\ \vert\ \H_0 \text{ is false})$ is the type II risk.
To estimate $\beta$, we need to repeat the same experiment multiple time and estimate the corresponding probability.
To estimate $\beta$, we need to repeat the same experiment multiple time and estimate the corresponding probability.
...
@@ -690,7 +690,7 @@ Decreasing $\alpha$ to reduce the type I error decreases the power of the test.
...
@@ -690,7 +690,7 @@ Decreasing $\alpha$ to reduce the type I error decreases the power of the test.
**Important:**
**Important:**
- confirm a detected effect with additional experiments/studies
- confirm a detected effect with additional experiments/studies
- the more (independent) studies, the lower risk of incorrect conclusion
- the more (independent) studies, the lower risk of incorrect conclusions
Eventually, we can also compare the number of significant SNPs found before and after p-value correction depending on the chosen type I risk alpha.
Eventually, we can also compare the number of significant SNPs found before and after p-value correction depending on the chosen type I risk $\alpha$.
```{r}
```{r}
# compute the number of significant SNPs
# compute the number of significant SNPs
...
@@ -1839,4 +1839,4 @@ Write me!
...
@@ -1839,4 +1839,4 @@ Write me!
---
---
[The .Rmd file corresponding to this page is available here under the AGPL3 Licence](https://lbmc.gitbiopages.ens-lyon.fr/hub/formations/ens_m1_ml/Practical_b.Rmd)
[The .Rmd file corresponding to this page is available here under the AGPL3 Licence](https://lbmc.gitbiopages.ens-lyon.fr/hub/formations/ens_m1_ml/Practical_c.Rmd)