The yEvo Mutation Browser: Enhancing student understanding of experimental evolution and genomics through interactive data visualization

To address the challenge of interpreting whole-genome sequencing data in the high school "yEvo" experimental evolution program, the authors developed the yEvo Mutation Browser, an interactive R Shiny web tool that visualizes mutations and protein structures to help students and researchers understand genetic adaptation while enabling the comparison of new datasets with a shared repository of yeast evolution data.

The Gist

Imagine you are a high school biology teacher. You've just led your students through a fascinating experiment: they took tiny baker's yeast cells and put them in a "pressure cooker" environment (like adding caffeine or antifungal drugs) to see how they would evolve to survive. The students watched the yeast adapt, just like animals adapting to a harsh climate, but on a microscopic scale.

Now comes the tricky part. The students have a list of genetic "typos" (mutations) found in the surviving yeast. But looking at a spreadsheet of thousands of genetic codes is like trying to find a specific needle in a haystack while wearing blindfolded. The students ask: "Which of these typos actually helped the yeast survive? Which ones were just random accidents?"

This is where the yEvo Mutation Browser comes in. Think of it as a high-tech, interactive detective board that turns that confusing spreadsheet into a clear, colorful story.

Here is how the tool works, explained through simple analogies:

1. The "City Map" (Chromosome Plot)

Imagine the yeast's DNA as a city with 16 different neighborhoods (chromosomes).

• Before: Students had a list of addresses where "crimes" (mutations) happened, but they couldn't see the city layout.
• Now: The Browser shows a giant map of the city. When a student clicks on a neighborhood, they see pins dropped where mutations happened.
• The Magic: If a pin is red, it means many different groups of students found a mutation in that exact spot. If it's white, it was just a one-time fluke. This instantly tells the students: "Hey, look! Everyone found a mutation in the 'PDR1' neighborhood. That's probably the key to surviving the caffeine!" It turns a guess into a pattern.

2. The "Pie Chart" (Mutation Types)

Imagine you are sorting a bag of mixed Legos.

• Before: Students just saw a list of broken pieces.
• Now: The Browser sorts them into a pie chart. It shows that most of the "survival" mutations are Missense mutations (pieces that changed the shape of the Lego slightly).
• The Lesson: This teaches students that while mutations happen randomly everywhere, the ones that actually help the yeast survive usually happen in the parts of the DNA that build proteins (the "active" parts of the cell).

3. The "Lollipop" (Gene View)

Imagine a gene is a long string of beads (amino acids).

• Before: Students couldn't tell where on the string the mutation happened.
• Now: The Browser draws a "lollipop chart." The stick is the gene, and the candy on top is the mutation.
• The Magic: If you see a cluster of lollipops all stuck in the same spot on the string, it's a huge clue. It suggests that changing that specific spot is what gave the yeast its superpower. It's like finding that every time a car breaks down, it's always the same bolt that is loose.

4. The "3D Sculpture" (Protein View)

This is the coolest part. Genes are instructions, but proteins are the actual machines built from those instructions.

• Before: Students saw a flat list of instructions.
• Now: The Browser builds a 3D hologram of the protein machine.
• The Magic: You can rotate the 3D model and see exactly where the mutations are. Are they on the outside of the machine? Are they jamming the gears? This helps students understand how the mutation changed the machine's function. It turns abstract code into a tangible object they can spin and inspect.

Why is this a big deal?

• For Students: It stops them from feeling overwhelmed by data. Instead of staring at a boring list, they get to be "data detectives," spotting patterns and solving the mystery of evolution themselves.
• For Teachers: It connects the classroom to the real world. The tool includes data from thousands of other students across the country. A student in Seattle can see how their results compare to a student in New York.
• For Scientists: It's not just for kids. Researchers can upload their own data to the same tool to see how their experiments stack up against years of previous yeast evolution studies.

The Bottom Line

The yEvo Mutation Browser is like giving students a pair of super-vision glasses. Without it, evolution data looks like a chaotic mess of static. With it, the static clears, and the beautiful, logical patterns of life's adaptation become crystal clear. It transforms a difficult homework assignment into an exciting discovery mission.

Technical Summary

1. Problem Statement

The yEvo (yeast evolution) project successfully engages high school students in experimental evolution by having them evolve Saccharomyces cerevisiae under selective pressures (e.g., caffeine, clotrimazole). While the hands-on laboratory component is effective, the downstream data analysis phase presents significant pedagogical challenges:

• Data Interpretation Barrier: Students struggle to interpret whole-genome sequencing (WGS) results and connect mutated genes to phenotypic adaptations.
• Lack of Context: Traditional analysis involves reviewing static lists of mutated genes and conducting independent literature searches. Students lack the broader context to distinguish between causative mutations and random "hitchhiker" mutations because they only see their own group's data without comparison to the larger population.
• Engagement Drop-off: The transition from wet-lab experiments to dry-lab bioinformatics often feels less interactive, leading to disengagement.

2. Methodology

A. Data Generation and Processing Pipeline

• Experimental Design: High school students evolved yeast under various stressors. University collaborators isolated clones, performed Illumina whole-genome sequencing, and analyzed the data.
• Variant Calling Pipeline:
  • Alignment: Reads aligned to the S. cerevisiae S288C reference genome using BWA.
  • Variant Calling: Three callers used: gatk4, lofreq, and freebayes.
  • Filtering: Quality filters applied via bcftools; ancestral variants removed using bedtools.
  • Annotation: Custom Python scripts annotated variants (coding, non-coding, transposon insertions). Transposon insertions were called using McClintock.
  • Validation: Manual review in Integrative Genomics Viewer (IGV) confirmed high-quality calls.
• Database Construction: All cumulative data since 2018 was compiled into a CSV file (the "yEvo Cumulative Database") containing over 2,000 genomic variants.

B. Software Development (The yEvo Mutation Browser)

• Framework: Built using R Shiny for the user interface and backend logic.
• Deployment: Hosted on shinyapps.io for web access; source code is open-source on GitHub.
• Architecture:
  • Session Isolation: Each user session loads an isolated copy of the cumulative database.
  • Data Upload: Users can upload local CSV/VCF data. Uploaded data is appended only to the user's session memory, allowing comparison with the cumulative dataset without altering the master database.
• External Integrations:
  • AllianceMine API: Retrieves gene metadata (SGD IDs, pathways).
  • UniProt Mapping API: Maps yeast genes to UniProt IDs for structural data.
  • Mol (MolStar) API:* Renders 3D protein structures (AlphaFold-predicted) directly in the browser, allowing interactive visualization of mutations on protein domains.
• AI Assistance: Large Language Models (Claude Code, ChatGPT) were used for code refinement and manuscript editing, with all outputs manually verified by authors.

3. Key Contributions: The Visualization Modules

The browser offers five interactive visualization modes to transform raw variant data into educational insights:

1. Chromosome Map:
   • Linear representation of the 16 nuclear chromosomes and mitochondrial genome.
   • Displays mutation locations as tick marks.
   • Recurrent Frequency: Color-codes genes based on how frequently they are mutated across the cumulative dataset (white = unique to one sample; red = highly recurrent). This helps students identify likely causative genes.
2. Variant Pie Chart:
   • Visualizes the distribution of mutation types (missense, nonsense, synonymous, indels, intergenic, etc.).
   • Highlights that phenotypic adaptation is primarily driven by coding sequence changes (missense mutations).
3. Mutation Spectrum:
   • Displays the ratio of Single Nucleotide Polymorphisms (SNPs) vs. Indels.
   • Distinguishes between transitions (more common) and transversions, helping students understand biochemical mutation biases.
4. Gene View (Lollipop Plot):
   • Linear plot showing mutation positions along the amino acid sequence and 5' upstream regions.
   • Hotspot Identification: The height of "lollipops" indicates mutation frequency at specific sites.
   • Includes a "Learn about gene" button linking to the Saccharomyces Genome Database (SGD).
5. Protein View:
   • Renders the 3D AlphaFold-predicted structure of the protein.
   • Maps mutations onto the 3D structure and linear domain tracks (Pfam).
   • Allows users to visualize if mutations cluster in specific structural domains or surface regions, aiding in hypotheses about Gain-of-Function (GOF) vs. Loss-of-Function (LOF).

4. Results and Educational Applications

• Pattern Recognition: The tool successfully allows students to distinguish between causative mutations and noise.
  • Example (Caffeine): Students observed that PDR1 and SKY1 were recurrently mutated across many samples, whereas other genes were unique to single groups. This confirmed PDR1 and SKY1 as key drivers of caffeine tolerance.
  • Example (Caspofungin): Analysis of FKS1 showed clustered missense mutations (suggesting GOF), while GSC2 showed scattered nonsense/frameshift mutations (suggesting LOF). This helped students hypothesize distinct mechanisms for drug resistance.
• Strain Background Comparison: The browser was used to compare a standard lab strain (S288C) against a mutator strain (msh2Δ, lacking mismatch repair).
  • Findings: The mutator strain showed a massive increase in total mutations (hitchhikers) and a bias toward indels. However, the core adaptive mutations (FKS1, GSC2) remained consistent, demonstrating that while mutator strains accelerate mutation accumulation, they complicate the identification of causative variants.
• User Adoption: The tool was deployed in all 2024–2025 yEvo classrooms. Teacher and student feedback was integrated into the design, resulting in a user-friendly interface that bridges the gap between raw data and biological insight.

5. Significance

• Educational Impact: The browser transforms abstract genomic data into tangible, visual patterns. It enables students to practice scientific inquiry by formulating hypotheses about mutation mechanisms (GOF vs. LOF) based on data patterns rather than just reading lists.
• Scalability and Reusability:
  • The tool is not limited to yEvo; non-affiliated researchers can upload their own experimental evolution or genetic screen data to compare against the yEvo cumulative dataset.
  • The open-source code allows other labs to instantiate their own browsers for different organisms or datasets.
• Future Directions: The authors plan to expand the tool to visualize Copy Number Variants (CNVs) and Structural Variants (SVs), which are currently missing but critical for a complete genomic picture.
• Community Resource: By aggregating data from multiple classrooms and years, the browser creates a shared, reproducible dataset that enriches the broader yeast genetics community and strengthens the link between K-12 education and professional research.