Skip to content

Course Description for Biometrics

Title: Bioinformatics

Duration: 14 weeks
Prerequisites: Introductory biology (molecular biology or genetics), one semester of programming (Python preferred), basic statistics
Target Audience: Upper-division undergraduates, graduate students, and working professionals seeking to expand their bioinformatics skill set
Modality: In-person or hybrid with hands-on lab sessions or fully independent self-paced study

Course Overview

This course provides a comprehensive introduction to bioinformatics with a distinctive emphasis on graph-based representations of biological data. Students learn how biological relationships — protein interactions, metabolic pathways, gene regulatory networks, evolutionary trees, and knowledge graphs — are naturally modeled as graphs, and how graph algorithms and databases unlock insights that tabular or sequence-only approaches miss.

The course moves from foundational sequence analysis through structural and systems biology, arriving at modern graph-powered approaches to drug discovery, precision medicine, and multi-omics integration. Each module pairs biological concepts with hands-on computational labs using Python, NetworkX, Neo4j, and visualization tools. A semester-long capstone project gives students experience solving a real bioinformatics problem end-to-end.

Topics Covered

Module 1 — Foundations (Weeks 1–3)

  • Week 1: Introduction to Bioinformatics and Biological Databases

    • Canonical data types in bioinformatics (sequences, structures, interactions, ontologies)
    • Major databases: NCBI, UniProt, PDB, Ensembl, KEGG, Reactome
    • Data formats: FASTA, FASTQ, GenBank, PDB, GFF3, OWL

  • Week 2: Introduction to Graph Theory for Biology

    • Nodes, edges, directed vs. undirected graphs, weighted graphs, bipartite graphs, labeled property graphs (LPGs)
    • Graph properties: degree distribution, clustering coefficient, centrality, connected components
    • Why biological data is inherently graph-structured
    • Labeled property graph model vs. RDF triple model
    • Scaling graph databases to billions of vertices with distributed graphs

  • Week 3: Graph Databases and Query Languages

    • Relational vs. graph databases for biological data
    • The Cypher query language
    • The GQL query language
    • Loading biological datasets into a graph database
    • An LPG graph data model for protein interaction
    • Hands-on lab: Running graph queries on a protein interaction graph

Module 2 — Sequence Analysis (Weeks 4–5)

  • Week 4: Sequence Alignment and Homology

    • Pairwise alignment: Smith-Waterman, Needleman-Wunsch
    • BLAST and sequence similarity searching
    • Scoring matrices (BLOSUM, PAM) and gap penalties
    • Sequence similarity networks as graphs
    • Graph data model for sequence similarity networks

  • Week 5: Phylogenetics and Evolutionary Graphs

    • Multiple sequence alignment (Clustal, MUSCLE)
    • Tree-building methods: neighbor-joining, maximum likelihood, Bayesian inference
    • Trees as directed acyclic graphs
    • Phylogenetic network models for reticulate evolution (horizontal gene transfer, hybridization)
    • Graph data model for evolutionary relationships

Module 3 — Structural Bioinformatics (Weeks 6–7)

  • Week 6: Protein Structure Prediction and Analysis

    • Levels of protein structure
    • Structure prediction: homology modeling, AlphaFold overview
    • A graph data model for protein contact maps
    • Residue interaction networks

  • Week 7: Molecular Interaction Networks

    • Protein-protein interaction (PPI) networks: yeast two-hybrid, co-immunoprecipitation, mass spectrometry
    • PPI databases: STRING, BioGRID, IntAct
    • Network topology analysis: hubs, bottlenecks, modules
    • Graph data model for PPI networks

Module 4 — Genomics and Transcriptomics (Weeks 8–9)

  • Week 8: Genome Assembly and Variation Graphs

    • De Bruijn graphs for genome assembly
    • Reference genome limitations and bias
    • Pangenome graphs and variation graphs (vg toolkit)
    • Graph data model for genomic variants

  • Week 9: Gene Regulatory Networks

    • Transcription factor binding, promoters, enhancers
    • RNA-seq analysis pipeline overview
    • Co-expression networks and differential network analysis
    • Inferring regulatory networks: WGCNA, ARACNE, GENIE3
    • Graph data model for gene regulation

Module 5 — Pathway and Systems Biology (Weeks 10–11)

  • Week 10: Metabolic Pathway Modeling

    • Metabolic networks as bipartite graphs (metabolites and reactions)
    • KEGG, Reactome, and BioCyc pathway databases
    • Flux balance analysis and constraint-based modeling
    • Graph data model for metabolic pathways

  • Week 11: Signaling Networks and Disease Modules

    • Cell signaling cascades as directed graphs
    • Network medicine: disease modules, network proximity, guilt by association
    • Drug-target-disease knowledge graphs
    • Graph data model for drug repurposing

Module 6 — Advanced Graph Applications (Weeks 12–13)

  • Week 12: Biomedical Knowledge Graphs and Ontologies

    • Gene Ontology (GO), Disease Ontology, Human Phenotype Ontology
    • Building knowledge graphs from heterogeneous biomedical data
    • Graph embeddings and link prediction for biological discovery
    • Graph neural networks (GNNs) for molecular property prediction
    • Graph data model for biomedical knowledge graphs

  • Week 13: Multi-Omics Integration and Graph Analytics at Scale

    • Integrating genomics, transcriptomics, proteomics, and metabolomics in a unified graph
    • Community detection algorithms for biological module discovery
    • Graph visualization examples: Vis-network, Cytoscape, Gephi
    • Graph data model for multi-omics integration
    • Scalability: graph partitioning, distributed graph databases
    • Appendix: A vis-network tutorial for informatics professionals

Module 7 — Capstone (Week 14)

  • Week 14: Capstone Presentations and Course Synthesis
    • Student capstone project presentations
    • Peer review and discussion
    • Future directions in graph-based bioinformatics

Topics Not Covered

This course does not include:

  • Wet-lab techniques — no bench work; the course is entirely computational
  • Clinical bioinformatics and HIPAA compliance — clinical data governance is mentioned but not taught in depth
  • Deep learning architectures in detail — GNNs are introduced conceptually but building custom deep learning models is beyond scope
  • Population genetics and GWAS — mentioned in the context of variation graphs but not covered as a standalone topic
  • Metagenomics and microbiome analysis — a natural extension but deferred to a follow-on course
  • R/Bioconductor programming — the course uses Python exclusively; R users can transfer concepts independently
  • Quantum computing for bioinformatics — too nascent for a survey course

Case Studies

Case Study 1: SARS-CoV-2 Variant Tracking with Phylogenetic Networks

Students build a phylogenetic network of SARS-CoV-2 spike protein sequences, identify recombination events that violate a strict tree model, and visualize variant lineage relationships as a graph in Neo4j.

Case Study 2: Drug Repurposing with a Biomedical Knowledge Graph

Using a simplified version of the Hetionet knowledge graph, students query drug-gene-disease relationships to identify candidate drugs for repurposing against a rare disease, applying network proximity and link prediction.

Case Study 3: Pangenome Graph for Structural Variant Discovery

Students construct a pangenome variation graph from a set of human genome assemblies, map short reads to the graph, and identify structural variants missed by traditional linear reference approaches.

Case Study 4: Cancer Driver Gene Identification via PPI Network Analysis

Students load a cancer-specific protein interaction network into Neo4j, compute centrality measures, run community detection, and cross-reference results with known cancer gene databases (COSMIC, OncoKB) to rank candidate driver genes.

Capstone Projects

Students select one of the following capstone projects (or propose their own with instructor approval). Each project requires a graph data model, a working implementation, a written report, and a 15-minute presentation.

Project 1: Antibiotic Resistance Gene Network

Build a graph database linking antibiotic resistance genes, mobile genetic elements, bacterial species, and antibiotics. Use graph queries to identify potential horizontal gene transfer pathways and predict resistance spread.

Graph data model: Nodes represent resistance genes, mobile genetic elements (plasmids, transposons), bacterial species, and antibiotic compounds. Edges capture "carried-by," "confers-resistance-to," "transferred-via," and "found-in-species" relationships with properties for evidence type and confidence score.

Project 2: Rare Disease Diagnosis Assistant

Construct a knowledge graph connecting phenotypes (HPO terms), genes, variants, and diseases. Given a set of patient phenotypes, use graph traversal and scoring to rank candidate diagnoses.

Graph data model: Nodes represent phenotypes (HPO), genes, genetic variants, and diseases (OMIM/Orphanet). Edges capture "associated-with," "causes," "variant-of," and "phenotype-of" relationships with properties for frequency, severity, and source database.

Project 3: Metabolic Model Comparison Across Organisms

Load genome-scale metabolic models for three related organisms into a graph database. Compare pathway topology, identify conserved and organism-specific modules, and visualize metabolic differences.

Graph data model: A bipartite graph where metabolite nodes and reaction nodes are connected by "substrate-of" and "product-of" edges. Reaction nodes link to enzyme and gene nodes. An "organism" property on each reaction enables cross-species filtering and comparison.

Project 4: Protein Function Prediction with Graph Embeddings

Generate graph embeddings from a protein interaction network and use them as features in a classifier to predict Gene Ontology annotations for unannotated proteins. Evaluate against held-out GO labels.

Graph data model: Protein nodes with properties for species, sequence length, and known GO annotations. Interaction edges carry experimental method, confidence score, and source database. GO term nodes form a separate DAG subgraph connected to proteins via "annotated-with" edges.

Project 5: Multi-Omics Patient Stratification

Integrate transcriptomic and proteomic data from a cancer cohort into a patient similarity network. Apply community detection to identify patient subgroups and correlate with clinical outcomes.

Graph data model: Patient nodes with clinical metadata (stage, survival, treatment). Gene and protein nodes connect to patients via weighted "expressed-in" edges. Patient-to-patient similarity edges are computed from multi-omics profiles with a similarity score property used for community detection.

Project 6: Custom Project (Instructor Approved)

Students may propose a bioinformatics problem of personal or professional interest that requires graph-based data modeling and analysis. A one-page proposal is due by Week 5.

Graph data model: The proposal must include a labeled property graph schema diagram showing node types, relationship types, and key properties, along with a justification for why a graph model is the appropriate representation for the chosen problem.

Assessment

Component Weight
Weekly labs and homework 30%
Midterm exam (Weeks 1–7) 15%
Case study reports (4 total) 20%
Capstone project 25%
Class participation and peer review 10%

Required Software and Tools

  • Python 3.10+ with Biopython, NetworkX, pandas, scikit-learn
  • Neo4j Desktop (Community Edition) or similar tool (Memgraph etc) that runs Cypher
  • Vis-network JavasScript library for network visualization
  • Jupyter Notebooks
  • Command-line tools: BLAST+, MUSCLE, vg toolkit

Learning Objectives by Bloom's Taxonomy

Level 1 — Remember

  1. List the major biological databases (NCBI, UniProt, PDB, KEGG, Reactome) and the types of data each contains
  2. Define fundamental graph theory terms: node, edge, degree, path, connected component, centrality, clustering coefficient
  3. Identify the standard bioinformatics file formats (FASTA, FASTQ, GenBank, PDB, GFF3)
  4. Recall the key steps in a basic RNA-seq analysis pipeline
  5. Name the three main approaches to phylogenetic tree construction (distance-based, maximum likelihood, Bayesian)

Level 2 — Understand

  1. Explain why biological relationships (interactions, pathways, regulation, evolution) are naturally represented as graphs rather than tables
  2. Describe the difference between a labeled property graph model and an RDF triple model
  3. Summarize how de Bruijn graphs are used in genome assembly
  4. Interpret degree distributions, hub-and-spoke patterns, and scale-free properties in biological networks
  5. Distinguish between sequence similarity networks, protein interaction networks, and gene regulatory networks in terms of what nodes and edges represent

Level 3 — Apply

  1. Perform pairwise and multiple sequence alignment using standard tools (BLAST, MUSCLE)
  2. Write Cypher queries to traverse and filter a biological graph database in Neo4j
  3. Load biological datasets from public databases into a graph database with an appropriate schema
  4. Compute network centrality measures (degree, betweenness, closeness) for a protein interaction network using NetworkX
  5. Construct a phylogenetic tree from a set of homologous sequences and interpret branch support values

Level 4 — Analyze

  1. Analyze the topology of a biological network to identify hubs, bottlenecks, and densely connected modules
  2. Compare graph-based pangenome representations with linear reference genomes for variant calling accuracy
  3. Evaluate the quality and completeness of a protein interaction network by assessing data source reliability and coverage bias
  4. Differentiate between correlation-based and information-theoretic methods for gene regulatory network inference
  5. Assess the strengths and limitations of different community detection algorithms when applied to biological networks

Level 5 — Evaluate

  1. Critically evaluate a published knowledge graph study for biological validity, data integration quality, and methodological rigor
  2. Judge the suitability of different graph database technologies (Neo4j, RDF stores, in-memory graph libraries) for a given bioinformatics problem
  3. Appraise the predictive value of graph embeddings and link prediction for drug repurposing compared with traditional approaches
  4. Evaluate trade-offs between network complexity, interoperability, and computational cost when modeling multi-omics data as a graph
  5. Assess when a graph-based approach adds genuine value versus when simpler tabular methods suffice

Level 6 — Create

  1. Design a graph data model for a novel bioinformatics dataset that captures entities, relationships, and properties appropriate to the biological domain
  2. Build an end-to-end bioinformatics pipeline that ingests raw data, constructs a graph, runs analytical queries, and produces a visualization
  3. Develop a knowledge graph that integrates data from at least three heterogeneous biomedical sources
  4. Construct a capstone project that formulates a biological question, selects appropriate graph methods, implements a solution, and communicates results
  5. Propose and justify a graph-based approach to an unsolved or open bioinformatics problem, including data model, algorithm selection, and validation strategy