PCA.Rmd

---
title: "Origin of anti-sickling activity via QSAR modelling"
author: "Chuleeporn Phanus-umporn, Saw Simeon, Watshara Shoombuatong and Chanin Nantasenamat"
date: "July 7, 2016"
output: pdf_document
---

## Principal Component Analysis 


```{r, error = FALSE, message = FALSE, warning=FALSE}
library(readxl)
library(caret)
library(dplyr)
library(ggplot2)
library(cowplot)


df <- read.csv("data.csv")
Activity <- df$Activity
descriptors <- df[, 2:ncol(df)]
data <- descriptors[, which(!apply(descriptors, 2, sd) == 0)]
raw <- cor(data)
df <- data
pca <- prcomp(df, retx=TRUE,scale.=TRUE, center = TRUE)
scores <- pca$x[,1:5]
loadings <- pca$rotation[,1:5]
```

### Explained variance, PC scores and loadings

```{r, error = FALSE, message=FALSE, warning=FALSE}
summary(pca)
scores
loadings

```

## PCA plot with clusters

```{r, error=FALSE, message=FALSE, warning=FALSE}
km <- kmeans(scores, center=5, nstart=5)
ggdata <- data.frame(scores, Cluster=km$cluster)
### paper numbering
library(grid)
set.seed(23)
x <- ggplot(ggdata, aes(x = PC1, y = PC2, colour = Cluster)) +
  geom_point(aes(fill=factor(Cluster)), size=5, shape=20, pch = 21, alpha = 0.8) +
  ggtitle(" ") +
  stat_ellipse(aes(fill=factor(Cluster)), colour = "black", 
               geom="polygon", level=0.95, alpha=0.2) +
  guides(color=guide_legend("Cluster"),fill=guide_legend("Cluster")) +
  #geom_text(aes(label=compoundnumber), size=7, hjust=0.5, vjust= 1.5, alpha=0.45) +
  theme(
    legend.position=("none"),
    #plot.title = element_text(size=20, face="bold", colour="black", vjust = 2, hjust=-0.07),
    panel.border = element_rect(linetype = "solid", colour = "black", fill = NA, size = 1),
    axis.text.y = element_text(size = 15),
    axis.ticks.length = unit(0.3, "cm"),
    axis.text.x = element_text(size = 15),
    legend.title=element_blank(),
    axis.title.x = element_text(color="black", size=20),
    axis.title.y = element_text(color="black", size=20)) +
  coord_cartesian(ylim = c(-15, 15), xlim = c(-15, 15))
x
```