OliFP.Rmd

---
title: "Predicting the Oligomerization of Fluorescent Protein"
author: "Rnewbie"
date: "February 22, 2558 BE"
output: pdf_document
---
```{r}
library(RCurl)
library(prospectr)
library(caret)
library(RWeka)
set.seed(3)
x <- getURL("https://raw.githubusercontent.com/Rnewbie/OliFP/master/OliFP.csv")
OliFP <- read.csv(text=x, header = TRUE)
data <- OliFP
PCP <- data[, 2:532]
DPC <- data[, 533:932]
AAC <- data[, 933:952]
AAC_DPC <- cbind(AAC, DPC)
AAC_PCP <- cbind(AAC, PCP)
DPC_PCP <- cbind(DPC, PCP)
ALL <- data[, 5:952]
Oligomerization <- data$Oligomerization
set.seed(1)
x <- list(AAC = AAC,
          DPC = DPC,
          PCP = PCP,
          AAC_DPC = AAC_DPC,
          AAC_PCP = AAC_PCP,
          DPC_PCP = DPC_PCP,
          ALL = ALL)
J48 <- lapply(x, function(x){
  data <- cbind(Oligomerization, x)
  Monomer <- subset(data, Oligomerization == "Monomer")
  Oligomer <- subset(data, Oligomerization == "Oligomer")
  sel <- kenStone(Monomer[-1], k = 150, metric = "mahal", pc=2)
  train_Monomer <- Monomer[sel$model, ]
  test_Monomer <- Monomer[sel$test, ]
  sel <- kenStone(Oligomer[-1], k = 149, metric = "mahal", pc=2)
  train_Oligomer <- Oligomer[sel$model, ]
  test_Oligomer <- Oligomer[sel$test, ]
  Train <- rbind(train_Monomer, train_Oligomer)
  Test <- rbind(test_Monomer, test_Oligomer)
  x <- J48(Oligomerization~., data = Train)
  cv <- evaluate_Weka_classifier(x,
                                         numFolds=10,
                                         complexity = FALSE,
                                         seed=1,
                                         class=TRUE)
  external <- evaluate_Weka_classifier(x,
                                       newdata = Test,
                                       numFolds=10,
                                       complexity = FALSE,
                                       seed=1,
                                       class=TRUE)
  Model <- summary(x)
  results <- list(Training = Model, Cross_Validation_10_Fold = cv, Testing = external)
  return(results)
})
J48
```
Performance table of J48 for training, 10-fold cross validation and external validation. 
```{r}
library(knitr)
x <- getURL("https://raw.githubusercontent.com/Rnewbie/OliFP/master/OliFPJ48.csv")
OliFPJ48 <- read.csv(text=x, header = TRUE)
kable(OliFPJ48, align = 'c', format = "markdown")
```