Bioinformatics Data Exploration and Visualization: PPI Visualization App:

AIVIA Project Engine AI in HealthCare & Medicine
, , , , , , ,
1

Bioinformatics Data Exploration and Visualization

Objective

The primary objective of this project is to develop a comprehensive understanding of bioinformatics data analysis and visualization techniques using the Stanford BioSNAP PPI (Protein-Protein Interaction) dataset. You will focus on handling bioinformatics data in R, creating both static and interactive visualizations, and developing a web application for data presentation using R Shiny. This exploration is fundamental for understanding complex bioinformatics networks and improving data communication in scientific contexts.


Learning Outcomes

By completing this project, you will:

  • Gain proficiency in handling and analyzing bioinformatics datasets using R.
  • Develop advanced skills in data visualization, including static and interactive plots.
  • Acquire expertise in building web applications for scientific data presentation using R Shiny.
  • Understand protein-protein interaction networks and their significance in bioinformatics research.
  • Enhance your ability to communicate scientific findings effectively through visual tools.

Prerequisites and Theoretical Foundations

1. R Programming (Basic to Intermediate Level)

  • Data Structures: Vectors, lists, matrices, data frames.
  • Control Flow: If-else statements, loops (for, while), functions.
  • Packages and Libraries: Installing and loading packages, understanding documentation.
Click to view R code examples 
# Basic data structures
vector <- c(1, 2, 3)
list <- list(a = 1, b = "text")
matrix <- matrix(1:9, nrow = 3)
data_frame <- data.frame(id = 1:3, value = c(10, 20, 30))

# Control flow
for (i in 1:5) {
  print(i)
}

# Functions
add_numbers <- function(x, y) {
  return(x + y)
}
result <- add_numbers(5, 3)

2. Introductory Knowledge of Bioinformatics

  • Basic Biology Concepts: Proteins, genes, DNA, RNA.
  • Protein-Protein Interactions (PPIs):
    • Understanding how proteins interact within a cell.
    • Significance of PPIs in cellular processes and disease mechanisms.
  • Bioinformatics Data:
    • Types of data (sequence data, interaction data).
    • Common bioinformatics databases and resources.
Click to view bioinformatics concepts
  • Proteins: Large, complex molecules that play many critical roles in the body.
  • PPIs: Physical contacts with molecular docking between proteins that occur in a cell or in a living organism.
  • PPI Networks: Graphical representations where nodes represent proteins and edges represent interactions.

3. Familiarity with Data Visualization Concepts

  • Understanding of Visualization Types:
    • Scatter plots, bar charts, heatmaps, network graphs.
  • Aesthetics and Clarity:
    • Importance of clear labeling, appropriate color schemes, and data-ink ratio.
  • Interactive Visualization Principles:
    • User engagement, responsiveness, and interactivity.
Click to view visualization concepts
  • ggplot2 Grammar of Graphics: Understanding how to build plots layer by layer.
  • Aesthetics Mapping: Mapping data variables to visual properties (e.g., color, size, shape).
  • Interactive Elements: Sliders, drop-down menus, hover information.

Skills Gained

  • Data Manipulation with dplyr:
    • Filtering, selecting, summarizing, and arranging data.
  • Data Visualization with ggplot2:
    • Creating and customizing a variety of plots.
  • Web Application Development with R Shiny:
    • Building interactive user interfaces.
    • Reactive programming concepts.
  • UI/UX Design Principles for Scientific Applications:
    • Designing intuitive and user-friendly interfaces.
    • Enhancing user engagement with interactive elements.
  • Understanding of PPI Networks:
    • Analyzing network structures.
    • Identifying key proteins and interactions.

Tools Required

  • Programming Language: R (version 3.6 or higher recommended)
  • Integrated Development Environment (IDE):
    • RStudio: Provides a user-friendly interface for R programming.
  • Libraries and Packages:
    • dplyr: Data manipulation (install.packages("dplyr"))
    • ggplot2: Data visualization (install.packages("ggplot2"))
    • R Shiny: Web application framework (install.packages("shiny"))
    • igraph: Network analysis and visualization (install.packages("igraph"))
    • reshape2: Data reshaping for visualization (install.packages("reshape2"))
  • Datasets:

Steps and Tasks

1. Data Acquisition and Setup

Tasks:

  • Download the Stanford BioSNAP PPI Dataset:
    • Access the dataset and save it to your working directory.
  • Install Necessary R Packages:
    • Ensure all required libraries are installed and loaded.

Implementation:

# Install necessary packages
install.packages(c("dplyr", "ggplot2", "shiny", "igraph", "reshape2"))

# Load libraries
library(dplyr)
library(ggplot2)
library(shiny)
library(igraph)
library(reshape2)
Data Download Instructions
  • Visit the Stanford BioSNAP Datasets page.
  • Locate the Protein-Protein Interaction (PPI) dataset.
  • Download the dataset (e.g., protein_interactions.csv) and place it in your working directory.

2. Bioinformatics Data Exploration in R

Tasks:

  • Load the PPI Dataset:
    • Read the CSV file into a data frame.
  • Explore the Data Structure:
    • View the first few rows and summary statistics.
  • Data Cleaning:
    • Check for missing values and duplicates.
  • Basic Data Analysis:
    • Use dplyr functions to manipulate and summarize the data.
  • Understand the PPI Network Structure:
    • Analyze the distribution of protein interactions.

Implementation:

# Load PPI dataset
ppi_data <- read.csv("protein_interactions.csv", stringsAsFactors = FALSE)

# Explore the data
head(ppi_data)
str(ppi_data)
summary(ppi_data)

# Data cleaning
ppi_data <- ppi_data %>%
  distinct() %>%        # Remove duplicate rows
  na.omit()             # Remove rows with missing values

# Basic data analysis
# Count interactions per protein
interaction_counts <- ppi_data %>%
  gather(key = "protein_role", value = "protein", protein1, protein2) %>%
  group_by(protein) %>%
  summarise(interaction_count = n()) %>%
  arrange(desc(interaction_count))

# View top proteins with most interactions
head(interaction_counts, 10)
Explanation
  • gather(): Converts wide data to long format.
  • group_by() and summarise(): Aggregate data to compute interaction counts.
  • arrange(desc()): Sorts data in descending order.

3. Data Visualization with ggplot2

Tasks:

  • Create Static Visualizations:
    • Frequency plots, heatmaps, network graphs.
  • Experiment with Different Plot Types:
    • Identify the most effective ways to represent protein interactions.
  • Customize Plots:
    • Enhance readability and aesthetics using ggplot2 themes and scales.

Implementation:

# Frequency plot of protein interactions
ggplot(interaction_counts[1:20, ], aes(x = reorder(protein, -interaction_count), y = interaction_count)) +
  geom_bar(stat = "identity", fill = "steelblue") +
  theme_minimal() +
  coord_flip() +
  labs(title = "Top 20 Proteins by Interaction Count",
       x = "Protein",
       y = "Number of Interactions")

# Heatmap of protein interactions
# Create an interaction matrix
interaction_matrix <- table(ppi_data$protein1, ppi_data$protein2)
interaction_matrix_melt <- melt(interaction_matrix)

# Plot heatmap
ggplot(interaction_matrix_melt, aes(x = Var1, y = Var2, fill = value)) +
  geom_tile() +
  scale_fill_gradient(low = "white", high = "red") +
  theme_minimal() +
  labs(title = "Heatmap of Protein Interactions",
       x = "Protein 1",
       y = "Protein 2",
       fill = "Interaction\nCount") +
  theme(axis.text.x = element_text(angle = 90, hjust = 1))
Explanation
  • reorder(): Reorders factors based on interaction count.
  • coord_flip(): Flips the axes for better readability.
  • melt(): Converts a matrix into a long format suitable for ggplot2.
  • geom_tile(): Creates a heatmap.

4. Network Visualization with igraph

Tasks:

  • Create a Graph Object:
    • Use igraph to represent the PPI network.
  • Visualize the Network:
    • Plot the network graph.
  • Analyze Network Properties:
    • Compute basic network metrics like degree, betweenness centrality.

Implementation:

# Create graph object
ppi_graph <- graph_from_data_frame(d = ppi_data, directed = FALSE)

# Basic network plot
plot(ppi_graph, vertex.size = 5, vertex.label = NA, edge.color = "gray")

# Compute degree centrality
degree_centrality <- degree(ppi_graph)
top_degrees <- sort(degree_centrality, decreasing = TRUE)[1:10]
print(top_degrees)

# Visualize network highlighting high-degree proteins
V(ppi_graph)$size <- ifelse(degree_centrality > quantile(degree_centrality, 0.95), 8, 3)
V(ppi_graph)$color <- ifelse(degree_centrality > quantile(degree_centrality, 0.95), "red", "lightblue")
plot(ppi_graph, vertex.label = NA, edge.color = "gray")
Explanation
  • graph_from_data_frame(): Creates a graph object from edge data.
  • degree(): Calculates the degree of each vertex.
  • Vertex Properties: Adjust vertex size and color based on degree.

5. Interactive Web Application with R Shiny

Tasks:

  • Design the User Interface (UI):
    • Create input controls for user interaction.
    • Layout the UI elements logically.
  • Develop the Server Logic:
    • Implement reactive expressions to handle user inputs.
    • Render outputs such as plots and tables based on user selections.
  • Implement Features:
    • Dynamic filtering of protein interactions.
    • Interactive network graphs with zoom and hover capabilities.
    • User-driven data exploration tools.
  • Apply UI Design Principles:
    • Ensure the interface is intuitive and user-friendly.
    • Use appropriate color schemes and fonts for readability.

Implementation:

# app.R

library(shiny)
library(dplyr)
library(ggplot2)
library(igraph)
library(visNetwork)

# Load data (ensure ppi_data and ppi_graph are available)

ui <- fluidPage(
  titlePanel("Protein-Protein Interaction (PPI) Network Explorer"),
  sidebarLayout(
    sidebarPanel(
      selectInput("selected_protein", "Select Protein:", 
                  choices = unique(ppi_data$protein1),
                  selected = unique(ppi_data$protein1)[1]),
      sliderInput("degree_filter", "Minimum Interaction Count:",
                  min = 1, max = max(degree_centrality), value = 1)
    ),
    mainPanel(
      tabsetPanel(
        tabPanel("Network Graph", visNetworkOutput("network_plot")),
        tabPanel("Interaction Table", dataTableOutput("interaction_table"))
      )
    )
  )
)

server <- function(input, output) {
  filtered_data <- reactive({
    ppi_data %>%
      filter(protein1 == input$selected_protein | protein2 == input$selected_protein)
  })
  
  output$network_plot <- renderVisNetwork({
    subgraph_nodes <- names(which(degree_centrality >= input$degree_filter))
    subgraph <- induced_subgraph(ppi_graph, vids = subgraph_nodes)
    
    visIgraph(subgraph) %>%
      visOptions(highlightNearest = TRUE, nodesIdSelection = TRUE)
  })
  
  output$interaction_table <- renderDataTable({
    filtered_data()
  })
}

# Run the application
shinyApp(ui = ui, server = server)
Explanation
  • visNetwork: An R package for interactive network visualization.
  • Reactive Expressions: Update outputs based on user inputs.
  • tabsetPanel: Organizes outputs into tabs.

6. Application in Bioinformatics Research

Tasks:

  • Analyze Findings:
    • Interpret the visualizations and analyses performed.
  • Identify Key Proteins and Interactions:
    • Use the tools developed to pinpoint proteins of interest.
  • Discuss Impact:
    • Explain how these tools aid in understanding PPI networks.
    • Highlight potential implications in disease research or drug discovery.
  • Communicate Findings:
    • Prepare reports or presentations using the visualizations created.
    • Share insights with the scientific community.

Implementation:

  • Interpretation:
    • Examine high-degree proteins as potential hubs in the network.
    • Analyze clusters or communities within the network.
  • Documentation:
    • Write summaries of findings.
    • Include visualizations in reports.
Example Analysis
  • High-Degree Proteins: Proteins with many interactions may play critical roles in cellular functions.
  • Network Clusters: Groups of proteins that interact closely may represent functional modules.

7. Next Steps and Enhancements

Suggestions:

  • Integrate Additional Datasets:
    • Incorporate gene expression data or disease associations.
  • Enhance the Shiny App:
    • Add more interactive features like search functionality or pathway analysis.
  • Advanced Visualizations:
    • Use 3D network visualizations or animation.
  • Machine Learning Applications:
    • Apply clustering algorithms to identify modules in the network.
  • Collaborate and Share:
    • Deploy the Shiny app online for wider access.
    • Collaborate with bioinformatics researchers for real-world applications.

Code Snippets

Click to view

1. Environment Setup

# Install necessary packages
install.packages(c("dplyr", "ggplot2", "shiny", "igraph", "reshape2", "visNetwork"))

# Load libraries
library(dplyr)
library(ggplot2)
library(shiny)
library(igraph)
library(reshape2)
library(visNetwork)

2. Bioinformatics Data Exploration

# Load PPI dataset
ppi_data <- read.csv("protein_interactions.csv", stringsAsFactors = FALSE)

# Data cleaning and exploration
ppi_data <- ppi_data %>%
  distinct() %>%
  na.omit()

# Summarize interaction counts
interaction_counts <- ppi_data %>%
  gather(key = "protein_role", value = "protein", protein1, protein2) %>%
  group_by(protein) %>%
  summarise(interaction_count = n()) %>%
  arrange(desc(interaction_count))

# View top proteins
head(interaction_counts, 10)

3. Data Visualization with ggplot2

# Frequency plot
ggplot(interaction_counts[1:20, ], aes(x = reorder(protein, -interaction_count), y = interaction_count)) +
  geom_bar(stat = "identity", fill = "steelblue") +
  theme_minimal() +
  coord_flip() +
  labs(title = "Top 20 Proteins by Interaction Count",
       x = "Protein",
       y = "Number of Interactions")

# Heatmap
interaction_matrix <- table(ppi_data$protein1, ppi_data$protein2)
interaction_matrix_melt <- melt(interaction_matrix)
ggplot(interaction_matrix_melt, aes(x = Var1, y = Var2, fill = value)) +
  geom_tile() +
  scale_fill_gradient(low = "white", high = "red") +
  theme_minimal() +
  labs(title = "Heatmap of Protein Interactions",
       x = "Protein 1",
       y = "Protein 2",
       fill = "Interaction\nCount") +
  theme(axis.text.x = element_text(angle = 90, hjust = 1))

4. Network Visualization with igraph

# Create graph object
ppi_graph <- graph_from_data_frame(d = ppi_data, directed = FALSE)

# Compute degree centrality
degree_centrality <- degree(ppi_graph)

# Visualize network with high-degree proteins highlighted
V(ppi_graph)$size <- ifelse(degree_centrality > quantile(degree_centrality, 0.95), 8, 3)
V(ppi_graph)$color <- ifelse(degree_centrality > quantile(degree_centrality, 0.95), "red", "lightblue")
plot(ppi_graph, vertex.label = NA, edge.color = "gray")

5. Interactive Web Application with R Shiny

# app.R

library(shiny)
library(dplyr)
library(igraph)
library(visNetwork)

# Assume ppi_data and ppi_graph are already loaded and processed

ui <- fluidPage(
  titlePanel("Protein-Protein Interaction (PPI) Network Explorer"),
  sidebarLayout(
    sidebarPanel(
      selectInput("selected_protein", "Select Protein:", 
                  choices = unique(ppi_data$protein1),
                  selected = unique(ppi_data$protein1)[1]),
      sliderInput("degree_filter", "Minimum Interaction Count:",
                  min = 1, max = max(degree_centrality), value = 1)
    ),
    mainPanel(
      tabsetPanel(
        tabPanel("Network Graph", visNetworkOutput("network_plot")),
        tabPanel("Interaction Table", dataTableOutput("interaction_table"))
      )
    )
  )
)

server <- function(input, output) {
  filtered_data <- reactive({
    ppi_data %>%
      filter(protein1 == input$selected_protein | protein2 == input$selected_protein)
  })
  
  output$network_plot <- renderVisNetwork({
    subgraph_nodes <- names(which(degree_centrality >= input$degree_filter))
    subgraph <- induced_subgraph(ppi_graph, vids = subgraph_nodes)
    
    visIgraph(subgraph) %>%
      visOptions(highlightNearest = TRUE, nodesIdSelection = TRUE)
  })
  
  output$interaction_table <- renderDataTable({
    filtered_data()
  })
}

# Run the application
shinyApp(ui = ui, server = server)

Conclusion

In this project, you have:

  • Developed proficiency in handling bioinformatics datasets using R and dplyr.
  • Created advanced data visualizations using ggplot2 to represent complex protein interaction networks.
  • Built an interactive web application using R Shiny for dynamic data exploration and visualization.
  • Enhanced your understanding of protein-protein interactions and their significance in bioinformatics research.
  • Applied UI design principles to create an intuitive and user-friendly interface for scientific data presentation.

This foundational knowledge prepares you for more advanced topics in bioinformatics and data science, such as:

  • Network Biology: Exploring the structure and function of biological networks.
  • Systems Biology: Integrating biological data to understand complex systems.
  • Data Integration and Mining: Combining multiple data sources for comprehensive analysis.
  • Machine Learning Applications in Bioinformatics: Predicting protein functions or disease associations using computational methods.