README.Rmd

---
output: github_document
---
<p align="center">
  <a href="https://github.com/frahik/GFR">
    <img src="Logo.png" alt="Genomic Functional Regression Logo"/>
  </a>
  <h3 align="center">
    Genomic Functional Regression analysis in R | Development version `r packageVersion('GFR')` 
  </h4>
  
  <p align="center">
  <a href="https://www.gnu.org/licenses/lgpl-3.0">
    <img src="https://img.shields.io/badge/License-LGPL%20v3-blue.svg" alt="LGPL, Version 3.0">
  </a> <a href="http://www.repostatus.org/#wip">
    <img src="http://www.repostatus.org/badges/latest/wip.svg" alt="Status of the Repo:  Initial development is in progress, but there has not yet been a stable, usable release suitable for the public">
  </a> <a href="">
    <img src="http://cranlogs.r-pkg.org/badges/GFR" alt="Dowloads from the CRAN">
  </a> <a href="https://cran.r-project.org/package=GFR">
    <img src="http://www.r-pkg.org/badges/version-ago/GFR" alt="CRAN">
  </a>
  </p>

  <h4 align="center">
    [Last README update: `r format(Sys.Date())`]
  </h4>
  
</p>


----------


# Table of contents
  - [NEWS](#news)
  - [Instructions](#instructions)
    - [Installation](#install)
    - [Demostration examples](#demo)
      - [Web version](#shinyapp)
      - [Available data](#data)
      - [Fitting a model](#fit-model)
      - [Predictive model](#predictive-model)
      - [Example #1 of how generate automatically a linear predictor](#auto-ETA)
      - [Example #2 of how generate automatically a linear predictor](#auto-ETA2)
      - [Example of how generate manually a linear predictor](#HM-ETA)
  - [How to cite this package](#cite)
  - [Contributions](#contributions)
  - [Authors](#authors)

<h2 id="news">
  News of this version (`r packageVersion('GFR')`)
</h2>

Revision 13

 - The BFR function now can obtain the PCC (Percentage of Correct Classification) to the Ordinal or binary data (`BFR(..., response_type = 'ordinal')`).
 - `forceClean` parameter from `cleanDat()` has changed to `noConfirm`.

Revision 12

 - The package includes a better BFR shiny app (use `runBFRInterface()` ).

Revision 10

 - The package now inclues a shiny app.

Past revisions

 - Rename from BGFRA -> BFR -> GFR
 - Initial development is in progress, but there has not yet been a stable, usable release suitable for the public; this is a pre-release, be careful.

<h2 id="instructions">
  Instructions for proper implementation
</h2>


<h3 id="install">
  Installation
</h3>

To complete installation of dev version of GFR from GitHub, you have to install [Rtools Software](https://cran.r-project.org/bin/windows/Rtools/) and a few packages first.

```{r installation, eval = FALSE}
install.packages('devtools')
devtools::install_github('frahik/GFR')
```

<h3 id="demo">
  Demostration examples
</h3>

<h4 id="shinyapp">
  Package Web version
</h4>

The package includes a shiny app to very easy use of the package, includes the most essencial parts, to use this,

```{r shinyapp, eval = FALSE}
library(GFR)
runInterface()
```

<h4 id="data">
  Availabe data
</h4>

Three data sets are available inside the package, to use it use `data()` function,

```{r data}
rm(list = ls())
library(GFR)
data("Wheat_GFR")
head(Wheat_GFR) # Load from data Wheat_GFR
paste0('Dimension of the Bands matrix: ', dim(Wheat_Bands)) # Load from data Wheat_GFR
paste0('Number of wavelenths:', length(Wheat_Wavelengths)) # Load from data Wheat_GFR

data("WheatI_GFR")
head(WheatI_GFR) # Load from data WheatI_GFR
paste0('Dimension of the Bands matrix: ',dim(WheatI_Bands)) # Load from data WheatI_GFR
paste0('Number of wavelenths:',length(WheatI_Wavelengths)) # Load from data WheatI_GFR

data("Maize_GFR")
head(Maize_GFR) # Load from data Maize_GFR
paste0('Dimension of the Bands matrix: ',dim(Maize_Bands)) # Load from data Maize_GFR
paste0('Number of wavelenths:',length( Maize_Wavelengths)) # Load from data Maize_GFR
```

<h4 id="fit-model">
  Fitting a model
</h4>

To more simple way to fit a model is with one environment and one trait data,

```{r fit_model}
data("Wheat_GFR")
data <- Wheat_GFR[which(Wheat_GFR$Env == 'Drought'), ]

fm <- BFR(data, nIter = 1000, burnIn = 300, verbose = F)

plot(fm)
```

<h4 id="predictive-model">
  Predictive model with a K-Folds Cross-validation
</h4>

To do a predictive model with a cross-validation, only we need to provide a list object with the type and the number of crossvalidation to do, in the package is available two cross-validatin types, in the following code, we see the CV K-Folds,

```{r predictive_model}
data("Wheat_GFR")
data <- Wheat_GFR[which(Wheat_GFR$Env == 'Drought'), ]
Crossvalidation_list <- list(Type = 'KFold', nFolds = 3)

pm <- BFR(data, nIter = 1000, burnIn = 300, set_seed = 10, CrossValidation = Crossvalidation_list, verbose = F)
summary(pm)

boxplot(pm)
```

<h4 id="auto-ETA">
  Generate automatically a linear predictor (Only Multi-Environment example)
</h4>

For more advanced predictions, we also provide a eta generator, for the estimations, this function automatically prepare the data and do a lot of validations, this is the recomended way to do a proper analysis,

```{r auto_ETA}
library(GFR)
data("Wheat_GFR")
CrossV <- list(Type = 'KFold', nFolds = 3)
ETA2 <- ETAGenerate(Wheat_GFR, datasetID = 'Line', priorType = 'BayesB', Bands = Wheat_Bands,
                    Wavelengths = Wheat_Wavelengths, method = 'Alternative2', basisType = 'Bspline.Basis', nBasis = 21)


pm2 <- BFR(ETA = ETA2, data, nIter = 1000, burnIn = 300, set_seed = 10, CrossValidation = CrossV, verbose = F)
summary(pm2)

plot(pm2)
```

<h4 id="auto-ETA2">
  Generate automatically a linear predictor (Multi-Trait & Multi-Environment example)
</h4>

Also, you can add some special terms, but be carefull,

```{r auto_ETA2}
data("Maize_GFR")
CrossV <- list(Type = 'RandomPartition', NPartitions = 3, PTesting = .25)
ETA3 <- ETAGenerate(Maize_GFR, basisType = 'Bspline.Basis', Bands = Maize_Bands, Wavelengths = Maize_Wavelengths, priorType = 'BRR', method = 'Simple', nBasis = 21)
ETA3$Design
ETA3$Basis
ETA3$Prior
ETA3$Method
pm3 <- BFR(ETA = ETA3, data, nIter = 1000, burnIn = 300, set_seed = 10, CrossValidation = CrossV, verbose = F)
summary(pm3)
```


<h4 id="HM-ETA">
  Handmade linear predictor
</h4>

If you are a expert and know what are you doing, you can generate the ETA manually,

```{r HM_ETA}
CrossV <- list(Type = 'KFold', nFolds = 3)
ETA4 <- list(Env = list(X = model.matrix(~0+as.factor(Wheat_GFR$Env)), model = 'FIXED'),
             Line = list(X = model.matrix(~0+as.factor(Wheat_GFR$Line)), model = 'BRR'),
             Bands = list(X = Bspline.Basis(Wheat_Bands, Wheat_Wavelengths, nBasis = 23), model = 'BayesA'))
pm4 <- BFR(data = Wheat_GFR, ETA = ETA4, nIter = 1000, burnIn = 300, CrossValidation = CrossV, set_seed = 10, verbose = F)
summary(pm4)

plot(pm4, select = 'MSEP')
boxplot(pm4, select = 'MSEP')
```

<h4 id="autoclean">
  Clean the working directory
</h4>

If you don't need anymore the files that the `BFR()` function creates, you can clean the directory with the following function provided in the package, 
```{r cleaner}
cleanDat(noConfirm = TRUE)
```


<h2 id="cite">
  How to cite this package
</h2>


First option, by the article paper

(Comming soon)

Second option, by the manual package

(Comming soon)



<h2 id="contributions">
  Contributions
</h2>

If you have any suggestions or feedback, I would love to hear about it. Feel free to report new issues in [this link](https://github.com/frahik/GFR/issues/new), also if you want to request a feature/report a bug, or make a pull request if you can contribute.


<h2 id="authors">
  Authors
</h2>

  
 - Francisco Javier Luna-Vázquez (Author, Maintainer)
 - Osval Antonio Montesinos-López (Author)
 - Abelardo Montesinos-López (Author)
 - José Crossa (Author)
 - Gustavo de los campos (Author)