Welcome
Welcome to the Course Website for EN.580.428 Genomic Data Visualization!
As the primary mode through which analysts and audience members alike consume data, data visualization remains an important hypothesis generating and analytical technique in data-driven research to facilitate new discoveries. However, if done poorly, data visualization can also mislead, bias, and slow down progress. This hands-on course will cover the principles of perception and cognition relevant for data visualization and apply these principles to genomic data, including large-scale single-cell and spatially-resolved omics datasets, using the R statistical programming language. Students will be expected to complete class readings, create weekly data visualizations as homework assignments, and make a major class presentation.
Course Information
Course Staff: Prof. Jean Fan and Caleb Hallinan
Lectures: 8:00am-9:50am Monday, Wednesday, and Friday. See Canvas for location details.
Office Hours: 10:00am-10:50am Monday, Wednesday, and by request. See Canvas for location details.
Course Details
☞ see Course tabFeatured Visualizations
Interrogating Spatial Spot Cluster Differential Gene Expression with 10x Visium
In these panels, I am depicting the representation of a 10x visium dataset in latent tSNE-embedded space and over the original spatial slide coordinates. I...
Identifying a Cluster of Breast Granular Cells
In the top left of my figure, I am depicting both my clusters made by kmeans clustering with k=7 in PCA space (with my cluster...
Homework 3: Differentially Expressed Genes analysis
[description] Those panels present a comprehensive visualization of Cluster 0 and its association with the gene SFRP4 through a combination of UMAP, spatial, and gene...
Spatial Transcriptomics Reveals a Distinct Epithelial Cell Population Defined by ELF3 Expression: A Multi-Dimensional Analysis of the Cluster in Interest
1. Describe your figure briefly so we know what you are depicting. Write a description to convince me that your cluster interpretation is correct.
Homework 2 submission
[description] In my visualization, I use points as the geometric primitive, angle and color for visual channel. The x-axis represents the PCA loadings for each...
Comparison of Scaled and Unscaled PCA: Gene Mean Expression, Variance, and PC1 Loadings
1. What data types are you visualizing? I am visualizing quantitative data, which includes log-transformed mean expression (x-axis), log-transformed variance (y-axis), and PC1 loading values...
Dimensionality Reduction using PCA
Homework 2
HW1: Gene expression pattern for GNB1 and HES4
1. What data types are you visualizing? I am visualizing HES4 and GNB1’s spatial gene expression patterns for eevee.
10 Genes with the Highest Counts are Expressed Across Most Spots
1. What data types are you visualizing? For this data visualization of the Eevee spatial transcriptomic data, I visualized both categorical data, the 10 genes...
Generation of Heatmap Expressing Top 20 Genes Within Pikachu Dataset
1. What data types are you visualizing? Within the Pikachu dataset that was visualized, gene expression levels across multiple individual cells proved to be a...
HW1: gene expression scatterplot
1. What data types are you visualizing? Spatial data of each cell, i.e the location of the cell within the section of the image, which...
All Visualizations
Animating the tSNE embedding of CODEX spleen dataset
I thought it would be fun to animate the transition from normal space (our x/y positional coordinates) to the tSNE embedded space. Thus, after normalizing my data, I first performed...
Eevee Dataset Deconvolution and Cell-type Analysis
Visualization Summary In this visualization, I analyzed the eevee dataset for unique cell-types using deconvolution (through STdeconvolve) and visualized it using scatterbar. I found 9 distinct cell types (Figure A),...
Comparison of PCA and t-SNE Methods
Visualization Summary In this visualization, I analyzed the differences between linear and nonlinear dimensionality reduction to visualized cells in the eevee dataset. The animation transitions between Principal Component Analysis and...
Visualization of nonlinear-embedded gex data versus linear-embedded gex data
Here, I illustrate the effect of an embedding in either PC-embedded (linear) space or tSNE-embedded (nonlinear) space. As observed, the PC-embedded shape resembles a volcano plot, with prominent spot placement...
Comparing Gene Expression in Normalized and Transformed Data
1. Written Answer Question: What happens if I do or not not normalize and/or transform the gene expression data (e.g. log and/or scale) prior to dimensionality reduction? I chose to...
A Comparative Analysis of Deconvolution and Clustering in Eevee Dataset
Write a brief description of your figure so we know what you are visualizing.
Using gganimate to understand different preprocessing methods
This animation visualizes how different preprocessing methods (raw, log-transformed, scaled, and log-scaled) affect t-SNE dimensionality reduction in spatial transcriptomics data. Each frame represents a different transformation method, highlighting variations in...
Deconvolution of Eevee Dataset for Identification of the Epithelial Cell Type
In this study, I analyzed the Eevee spatial transcriptomics dataset using deconvolution techniques and clustering methods to identify distinct cell types and visualize their gene expression patterns. The dataset was...
Visualization of t-SNE on Different PC Counts on Pikachu Dataset
1.If I perform non-linear dimensionality reduction on PCs, what happens when I vary how many PCs should I use?
Difference between STDecovolution or K-mean clustering on Eevvee Dataset
### Same as homework 4, I performed STdeconvolve on the Eevvee dataset to infer cell-type proportions and used K-means clustering (K=7) to analyze tissue organization. The scatterbar plot visualizes the...
