report.Rmd

---
title: "Workflow results report"
output: 
  pdf_document:
    fig_crop: no
    fig_caption: no
params:
  input: NA
  original_data: NA
  classes: NA
  positives: NA
  negatives: NA
  port: NA
  original_pca_data: NA
  original_fisher_data: NA
  after_fisher_pca_data: NA
  after_fisher_data: NA
  after_relieff_pca_data: NA
  after_relieff_data: NA
  after_plsda_perf: NA
  after_splsda: NA
---
```{r, include=FALSE}
options(tinytex.verbose = FALSE)
library(knitr)
knit_engines$set(asis = function(options) {
  if (options$echo && options$eval) knit_child(text = options$code)
})
knitr::opts_chunk$set(echo = FALSE) 
```
## Initial state and Fisher filter
### Data
Number of variables: `r dim(params$original_data)[2]`\newline
Number of samples: `r dim(params$original_data)[1]`\newline
Classes: `r levels(factor(params$classes))` \newline
Number of samples of class "`r levels(factor(params$classes))[1]`": `r length(params$positives)`\newline
Number of samples of class "`r levels(factor(params$classes))[2]`": `r length(params$negatives)`\newline
Number of variables kept after Fisher: `r params$input$fisher_vars`\newline

### Plots

```{r}
plot(params$original_pca_data, main = 'Principal components with all variables')
```


```{r}
plot(params$original_fisher_data$to_plot, ylab = 'Score', main = 'Fisher scores')
abline(v = params$input$fisher_vars, col = 'red', lwd = 2)
```


```{r}
plot(params$after_fisher_pca_data, main = 'Principal components with the selected variables after Fisher scoring')
```


```{asis, echo=params$input$use_relieff, eval=params$input$use_relieff}
## ReliefF
### Data
Number of variables kept after ReliefF: `r params$input$relieff_vars`\newline
Number of nearest neighbors chosen: `r params$input$relieff_k`\newline
Number of iterations made: `r if (params$input$relieff_iters == 0) { params$input$fisher_vars } else if (params$input$relieff_iters == -1) { floor(log(params$input$fisher_vars)) } else { floor(sqrt(params$input$fisher_vars)) }`\newline
Algorithm chosen to calculate the scores: `r params$input$relieff_method`\newline

### Plots
```


```{r, eval=params$input$use_relieff}
plot(params$after_relieff_data$sorted_attrs, ylab = 'Score', main = 'ReliefF scores')
abline(v = params$input$relieff_vars, col = 'red', lwd = 2)
```


```{r, eval=(params$input$draw_relieff_heatmap && params$input$use_relieff)}
heatmap(params$after_relieff_data$data, main = paste('Heatmap of the', params$input$relieff_vars, 'variables after applying ReliefF'))
```


```{r, eval=params$input$use_relieff}
plot(params$after_relieff_pca_data, main = 'Principal components with the selected variables after ReliefF')
```

## sPLS-DA
### Data
Number of components to do the perfectioning: `r params$input$components`\newline
Number of cross validation folds performed: `r params$input$cv_folds`\newline
Number of cross validation repetitions performed: `r params$input$cv_repeats`\newline

### Plots

```{r}
matplot(params$after_plsda_perf$perf_plsda$error.rate$BER, type = 'l', lty = 1, col = color.mixo(1:3), main = 'Balanced Error Rate for amount of components', ylab = 'Balanced Error Rate')
legend('topright', c('max.dist', 'centroids.dist', 'mahalanobis.dist'), lty = 1, col = color.mixo(1:3))
```


```{r}
plot(params$after_plsda_perf$tune_splsda, col = color.jet(params$input$components), main = 'Error rates for number of components')
```


```{r, eval=params$input$draw_indiv_plot}
plotIndiv(params$after_splsda$splsda, comp = c(1, 2), ind.names = FALSE, legend = TRUE, ellipse = TRUE, title = 'sPLS-DA, final result, components 1 and 2')
```


```{r, eval=params$input$draw_auroc}
ncomp = params$after_plsda_perf$final_ncomp
for (i in seq(ncomp)){
  auroc(params$after_splsda$splsda, roc.comp = i)
}
```

```{asis, echo=params$input$draw_cim, eval=params$input$draw_cim}
![](TFG_workflow/`r paste("cim", params$port, ".png", sep = "_")`)
```


```{r, eval=params$input$draw_loadings}
ncomp = params$after_plsda_perf$final_ncomp
for (i in seq(ncomp)){
  plotLoadings(params$after_splsda$splsda, comp=i, contrib = 'max', method = 'mean', title = paste('Loadings of component', i))
}
```

### Selected variables

```{r}
vars_list = c()
ncomp = params$after_plsda_perf$final_ncomp
for (i in seq(ncomp)){
  new_vars = selectVar(params$after_splsda$splsda, comp=i)$name
  vars_list = append(vars_list, new_vars[which(!new_vars %in% vars_list)])
}
```

`r paste0(vars_list, collapse = "\\newline ")`