09_ExploreModelMatrix.Rmd

# Interpreting model coefficients with ExploreModelMatrix

Instructor: Leo

## Model objects in R

* Linear regression review https://lcolladotor.github.io/bioc_team_ds/helping-others.html#linear-regression-example
* With R, we use the `model.matrix()` to build regression models using the `Y ~ X1 + X2` formula syntax as exemplified below.

```{r model.matrix}
## ?model.matrix
mat <- with(trees, model.matrix(log(Volume) ~ log(Height) + log(Girth)))
mat
colnames(mat)
```

* How do we interpret the columns of our model matrix `mat`?

```{r lm_example}
summary(lm(log(Volume) ~ log(Height) + log(Girth), data = trees))
```

## ExploreModelMatrix

* It's a Bioconductor package which is useful to understand statistical models we use in differential expression analyses. It is interactive and helps us by creating some visual aids.
  - http://www.bioconductor.org/packages/ExploreModelMatrix
* For more details, check their paper https://doi.org/10.12688/f1000research.24187.2.
* We'll go over the examples they provide at http://www.bioconductor.org/packages/release/bioc/vignettes/ExploreModelMatrix/inst/doc/ExploreModelMatrix.html


## Example 1

```{r EMM_example1}
## Load ExploreModelMatrix
library("ExploreModelMatrix")

## Example data
(sampleData <- data.frame(
    genotype = rep(c("A", "B"), each = 4),
    treatment = rep(c("ctrl", "trt"), 4)
))

## Let's make the visual aids provided by ExploreModelMatrix
vd <- ExploreModelMatrix::VisualizeDesign(
    sampleData = sampleData,
    designFormula = ~ genotype + treatment,
    textSizeFitted = 4
)

## Now lets plot these images
cowplot::plot_grid(plotlist = vd$plotlist)
```

Interactively, we can run the following code:

```{r EMM_example1_interactive, eval = FALSE}
## We are using shiny again here
app <- ExploreModelMatrix(
    sampleData = sampleData,
    designFormula = ~ genotype + treatment
)
if (interactive()) shiny::runApp(app)
```

## Example 2

http://bioconductor.org/packages/release/bioc/vignettes/ExploreModelMatrix/inst/doc/ExploreModelMatrix.html#example-2

## Example 3

http://bioconductor.org/packages/release/bioc/vignettes/ExploreModelMatrix/inst/doc/ExploreModelMatrix.html#example-3

## Exercise

<style>
p.exercise  {
background-color: #E4EDE2;
padding: 9px;
border: 1px solid black;
border-radius: 10px;
font-family: sans-serif;
}
</style>

<p class="exercise">
**Exercise 1**:
Interpret `ResponseResistant.Treatmentpre` from the second example. It could be useful to take a screenshot and to draw some annotations on it.
</p>

<p class="exercise">
**Exercise 2**:
Whis is the `0` important at the beginning of the formula in the third example?
</p>

## To learn more

_A guide to creating design matrices for gene expression experiments_:

* http://bioconductor.org/packages/release/workflows/vignettes/RNAseq123/inst/doc/designmatrices.html
* https://f1000research.com/articles/9-1444

_“Model matrix not full rank”_

* http://bioconductor.org/packages/release/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#model-matrix-not-full-rank

## Community

Some of the `ExploreModelMatrix` authors:

* https://twitter.com/CSoneson
* https://twitter.com/FedeBioinfo
* https://twitter.com/mikelove