🔸 A-DE-1: Biomedical Knowledge Graphs

Biomedical Knowledge Graphs

Advanced Track - Domain-Specific AI Explorations


Develop a comprehensive understanding of protein-protein interactions by constructing and analyzing a Knowledge Graph using the Stanford BIOSNAP Datasets. This project will focus on performing network analysis to recreate and examine Strongly Connected Components (SCC) and to identify significant clusters within the network. This exploration is fundamental for uncovering hidden patterns in the domain of bioinformatics.

Learning Outcomes

  • Develop expertise in constructing and analyzing Knowledge Graphs with a focus on biological data.
  • Acquire advanced skills in data visualization to effectively represent interaction networks.
  • Gain a thorough understanding of protein-protein interaction dynamics and their implications in disease modeling and drug discovery.

Pre-requisite Skills

  • Python or R programming (Intermediate level)
  • Basic understanding of graph theory
  • College-level biology to ensure a foundational understanding of the biological concepts related to protein interactions.

Skills Gained

  • Construction and analytical evaluation of Knowledge Graphs.
  • Application of sophisticated network analysis techniques.
  • Advanced data visualization to interpret complex biological datasets.

Tools Explored

  • NetworkX: For constructing and analyzing network graphs in Python.
  • Pandas: Essential for data manipulation in Python.
  • igraph: An R package for network analysis.
  • tidyverse: A collection of R packages for data manipulation and visualization.

Steps and Tasks

At any point during your project, if you find yourself needing assistance, several resources are available to support you:

  • Code Snippets: Code snippets for each step are shared in the next section.
  • Code-Along Discussions Category: Join discussions to exchange ideas and resolve issues.
  • STEM-Away Mentorship Category: For paid members, access live mentorship, including forum support and webinars, to enhance your learning experience.

1. Data Acquisition and Setup

  • Download the Stanford BIOSNAP Protein-Protein Interaction dataset.
  • Install and configure Python and R environments with necessary libraries (NetworkX, Pandas, igraph, tidyverse).

2. Knowledge Graph Construction

  • Use NetworkX or igraph to construct the knowledge graph from the interaction data.
  • Visualize the graph using tools available in Pandas and tidyverse.
  • Experiment with different visualization libraries to find the most effective way to represent your data. Consider using:
    • Matplotlib and Seaborn for basic plotting.
    • Plotly or Bokeh for interactive visualizations.

3. Network Analysis

  • Implement algorithms to identify and analyze Strongly Connected Components (SCC) within the graph.
  • Perform cluster analysis to identify key nodes and interactions.
  • Explore various graph analysis techniques to deepen your understanding of network dynamics:
    • Centrality Measures: Identify the most influential proteins in the network.
    • Community Detection: Discover clusters or communities within the protein interactions.
  • Consider using parallel computing or efficient data structures to handle large datasets and complex computations.

5. Application in Drug Discovery

After constructing and analyzing the knowledge graph, you can explore its application in drug discovery:

  • Incorporate External Drug Target Information: Use databases such as DrugBank, UniProt, or ChEMBL to gather information on known drug targets. Integrate this data into your graph by marking proteins that are known drug targets.
  • Identifying Potential Drug Targets: Use centrality measures to identify highly influential proteins in the network. Proteins with high centrality are often critical in various biological processes and can be potential targets for drug development.
  • Interaction Analysis for Drug Targeting: Analyze the interactions between potential drug target proteins and other proteins. Look for proteins that interact with multiple known drug targets, as these could be potential candidates for therapeutic interventions.

6. Disease Pathway Analysis Using External Annotations

To explore disease pathways:

  • Utilize Disease-Associated Protein Annotations: Gather annotations from databases like OMIM, GeneCards, or DisGeNET that link proteins to specific diseases.
  • Reconstruct Disease-Specific Subgraphs: Extract subgraphs based on these annotations to visualize and analyze pathways related to specific diseases. This can help identify key proteins and interactions in disease mechanisms.

Code Snippets

Click to view

The following code snippets provide a comprehensive guide to setting up your environment, preparing the data, constructing the knowledge graph, performing initial network analysis, and example application.

1. Environment Setup

Before you start, ensure all necessary libraries are installed to handle the dataset and perform graph-based operations. This setup includes libraries for data manipulation, graph construction, and visualization.

# Install NetworkX for graph operations, Pandas for data manipulation, and iGraph for advanced graph analysis
!pip install networkx pandas python-igraph

# Additional visualization libraries
!pip install matplotlib seaborn

2. Loading and Preparing the Data

Load the dataset using Pandas, a powerful tool for data analysis. This step includes preliminary data checks such as viewing the top rows of the dataset and summary statistics to understand the nature of the data.

import pandas as pd

# Load the dataset
data = pd.read_csv('protein_interactions.csv')

# Display the first few rows of the dataframe

# Get a summary of the dataframe

3. Constructing the Knowledge Graph

Use NetworkX to construct the knowledge graph from the interaction data. This involves creating a graph object and adding edges from the dataset, representing protein interactions.

import networkx as nx

# Create a graph from the Pandas dataframe
G = nx.from_pandas_edgelist(data, source='protein1', target='protein2')

# Display basic information about the graph

4. Basic Graph Analysis

Perform basic graph analysis to understand the structure and key metrics of the graph. This includes calculating the number of connected components and visualizing the network.

# Calculate the number of connected components
connected_components = nx.number_connected_components(G)
print("Number of connected components:", connected_components)

# Visualize the graph
import matplotlib.pyplot as plt

plt.figure(figsize=(12, 8))
nx.draw(G, node_color='lightblue', with_labels=True, node_size=500, font_size=10)
plt.title('Protein-Protein Interaction Graph')

5. Advanced Network Analysis

Implement more complex network analysis techniques such as calculating centrality measures to identify key nodes within the network.

# Calculate degree centrality of the graph
centrality = nx.degree_centrality(G)
sorted_centrality = sorted(centrality.items(), key=lambda x: x[1], reverse=True)

# Display the top 5 nodes with the highest centrality
print("Top 5 nodes by degree centrality:", sorted_centrality[:5])

6. Using iGraph for Advanced Analysis

For those interested in exploring more sophisticated analyses, convert the NetworkX graph to an iGraph object. iGraph provides powerful tools for community detection and network motifs.

from igraph import Graph

# Convert NetworkX graph to iGraph
ig = Graph.TupleList(G.edges(), directed=False)

# Perform community detection
community = ig.community_multilevel()
print("Modularity of found partitions:", community.modularity)

7. Applications

# Load disease association data
disease_associations = pd.read_csv('disease_associations.csv')

# Create subgraphs for specific diseases
disease_pathways = {}
for disease in disease_associations['disease'].unique():
    associated_proteins = disease_associations[disease_associations['disease'] == disease]['protein_id']
    subgraph = G.subgraph(associated_proteins)
    disease_pathways[disease] = subgraph

# Analyze a specific disease pathway
scc_disease = nx.strongly_connected_components(disease_pathways['Alzheimer’s Disease'])

8. Visualization Techniques

import plotly.graph_objs as go

# Create a trace for the nodes
node_trace = go.Scatter(
            title='Node Connections',

# Create a trace for the edges
edge_trace = go.Scatter(
    line=dict(color='rgb(125,125,125)', width=1),

for node in G.nodes():
    x, y = G.nodes[node]['pos']
    node_trace['hovertext'].append(f"Protein: {node}")

for edge in G.edges():
    x0, y0 = G.nodes[edge[0]]['pos']
    x1, y1 = G.nodes[edge[1]]['pos']
    edge_trace['x'] += [x0, x1, None]
    edge_trace['y'] += [y0, y1, None]

fig = go.Figure(data=[edge_trace, node_trace],
                    title='Protein-Protein Interaction Network',
                    margin=dict(b=20, l=5, r=5, t=40),
                    xaxis=dict(showgrid=False, zeroline=False, showticklabels=False),
                    yaxis=dict(showgrid=False, zeroline=False, showticklabels=False)


Self Evaluation for AI Mentor/Evaluator Conversation

Prepare for your discussions with the AI mentor by reflecting on the following points. These reflections will help you articulate your experiences, challenges, and insights effectively, showcasing not just your technical skills but also your problem-solving abilities, curiosity, and understanding of real-world applications.

For Virtual-Internships Admissions:

The points provided are designed to guide your conversation for admission into our Virtual-Internships program. Focus on demonstrating your enthusiasm for learning, your ability to tackle challenges, and your understanding of how this project relates to real-world applications. The AI mentor is looking for candidates who show potential, a growth mindset, and a genuine interest in the field. Remember, it’s not just about what you’ve done, but also about what you’ve learned and how you’ve grown through the process.

For Skill Certifications on the Talent Discovery Platform:

After completing your Virtual-Internship, you’ll be well-equipped to engage in more in-depth technical discussions. The points under this section are tailored for these advanced conversations, which are part of our Skill Certification process on the Talent Discovery Platform. Here, you’ll demonstrate not just proficiency in the tools and techniques, but a deep understanding of algorithmic choices, system design considerations, and even ethical implications of your work. These discussions are designed to showcase your readiness for high-level industry roles or advanced academic pursuits.

If you already have the required expertise to handle an advanced conversation, feel free to skip the virtual-internships and jump straight into skill certifications!


  • For both types of conversations, the AI mentor will adapt to your level of expertise. Be honest about your current knowledge and eager to learn.
  • In both cases, don’t just list what you did; explain why you made certain choices and what you learned from the outcomes.
  • The Virtual-Internship experience is designed to prepare you for the more rigorous Skill Certification evaluations. Embrace the learning process, and you’ll find yourself naturally growing into the depth required for these advanced discussions.
  • Last but definitely not the least, talk about soft skills wherever applicable!

Now, let’s dive into the specific points for each type of evaluation. Feel free to explore these topics in the order that best tells your unique project story:

Evaluation for our Virtual-Internships Admissions
  • Start with a Brief Project Overview: Begin by summarizing the project objectives and the key technologies you used (Knowledge Graphs, Stanford BIOSNAP Datasets, NetworkX, igraph). This sets the context for the discussion.

  • Discuss Graph Construction: Explain the process of constructing the knowledge graph. Discuss any challenges you faced, such as handling large-scale data or dealing with incomplete data, and how you addressed these issues.

  • Challenges and Problem-Solving: Present a specific challenge you faced, like identifying key nodes or handling noisy data. Explain your solution and how it impacted the graph’s quality and insights. This shows critical thinking and problem-solving skills.

  • Insights from Network Analysis: Share an interesting finding from your network analysis. For example, “I found that using centrality measures helped identify key proteins involved in specific biological processes, which could be potential targets for drug development.”

  • Real-world Application: Discuss how you would apply this knowledge graph in real-world scenarios. Talk about potential applications in drug discovery or disease modeling and the implications of your findings.

  • Learning and Growth: End by reflecting on your learning journey such as “Working on this project, I gained a deep appreciation for how graph structures can capture complex biological interactions. I also realized the importance of data preprocessing - our initial results improved significantly after we properly handled missing data.”

  • Ask Questions: Show curiosity by asking the AI mentor questions. For example, “I’m curious, how do large-scale bioinformatics platforms handle the integration of diverse biological data sources in constructing comprehensive knowledge graphs?”

Evaluations for Skill Certifications for our Talent Discovery Platform
  • Graph Construction and Completeness:

    • Graph Accuracy: Discuss the accuracy and completeness of the knowledge graph you constructed. Provide examples of how well the graph represented the dataset and any areas where it could be improved.
    • Challenges Faced: Describe any significant challenges encountered during the graph construction and how you addressed these issues.
  • Network Analysis Techniques:

    • Analysis Methods: Explain the different network analysis techniques you used, such as centrality measures or community detection. Highlight significant findings or challenges you encountered.
    • Learning Curves: Discuss the trends observed during the analysis process, highlighting any substantial discoveries or persistent challenges.
  • Scalability and Practical Applications:

    • Handling Complex Datasets: Describe any challenges you faced with the dataset’s complexity and size. Discuss strategies you employed for managing large-scale data and optimizing your graph analysis.
    • Application of Findings: Share how the insights gained from the network analysis could be applied in real-world scenarios, particularly in the domain of bioinformatics.
  • Comparative Analysis and Methodology Evaluation:

    • Methodology Comparison: Compare the methodologies used in constructing and analyzing the knowledge graph, such as different graph construction tools or analysis techniques. Highlight how certain methodologies were more effective in revealing biological insights.
    • Tool Effectiveness: Evaluate the effectiveness of the tools and libraries used, such as NetworkX versus igraph, or Pandas versus tidyverse for data manipulation and visualization.
  • Domain-Specific Considerations:

    • Protein-Protein Interactions (PPI): Discuss the importance of protein-protein interactions in understanding biological processes and disease mechanisms. Highlight how your analysis can contribute to identifying key proteins and potential drug targets.
    • Related Fields and Tools: Mention other relevant fields such as genomics, transcriptomics, or metabolomics, and how similar network analysis techniques can be applied. Discuss the use of databases like STRING, BioGRID, or DrugBank for gathering additional interaction data or validating your findings.
    • Advanced Techniques and Tools: Share any advanced techniques or tools you explored, such as using Cytoscape for network visualization or incorporating machine learning algorithms for predictive modeling in bioinformatics.

We recommend covering 3-5 different areas of the project. Remember, the goal is not just to showcase your technical skills but also to demonstrate your ability to think critically about the application, challenges, and real-world implications of your work. The AI chatbot will likely engage more deeply if you present a well-rounded perspective that goes beyond just coding.

Resources and Learning Materials

Online Resources:

  1. NetworkX Documentation: The official NetworkX documentation provides comprehensive guides, tutorials, and API references for working with graphs and networks in Python.
  2. igraph Documentation: The igraph documentation offers detailed information on using the igraph library for network analysis in R, including examples and explanations of various functions and algorithms.
  3. Cytoscape: Cytoscape is an open-source software platform for visualizing and analyzing molecular interaction networks, including protein-protein interaction networks. The website provides tutorials, user guides, and resources for getting started with Cytoscape.
  4. Biological General Repository for Interaction Datasets (BioGRID): BioGRID is a curated database of protein-protein interactions and genetic interactions. It can serve as a valuable resource for acquiring additional datasets or validating protein interaction data.
  5. STRING: The STRING database provides known and predicted protein-protein interactions, as well as functional associations between proteins. It can be a useful resource for exploring and understanding protein interaction networks.

Research Papers:

  1. Barabási, A. L., & Oltvai, Z. N. (2004). Network biology: understanding the cell’s functional organization. Nature Reviews Genetics, 5(2), 101-113. [Link] This paper provides an overview of network biology and the importance of understanding the functional organization of cellular networks, including protein-protein interaction networks.
  2. Vidal, M., Cusick, M. E., & Barabási, A. L. (2011). Interactome networks and human disease. Cell, 144(6), 986-998. [Link] This review article discusses the role of interactome networks (including protein-protein interaction networks) in understanding human diseases and their potential applications in drug discovery.

Access the Code-Along for this Skill-Builder Project to join discussions, utilize the t3 AI Mentor, and more.