Tyler Slonecki

Your home institution logo goes here

Clemson Research Experiences for Undergraduates

Collaborative Data Visualization Applications Summer 2014

Home Institution

Wofford College
Spartanburg, SC

Clemson Research Mentor

Dr. Sapna Sarupria
Molecular Modeling and Simulation

Clemson Visualization Mentor

Dr. Vetria Byrd
Director of Advanced Visualization

About Me

I'm a Mathematics and Biology double major with an Emphasis in Computational Science at Wofford College located in nearby Spartanburg, SC. Among many things, I enjoy being outdoors, running, playing sports, watching sports and doing anything involving physical activity. I also maintain a strong appreciation for music and play several instruments. Academically, I enjoy all aspects of my majors and I plan to pursue a career that involves Biology, Mathematics and Computer Science. Right now, my aim is to attend graduate school in Biomedical Engineering, which I feel would encompass all three of my educational areas. This summer I am being assisted financially by both the NSF through Clemson University and my home institution through the Brown Scholarship Fund.

Project Description

The Research Background

This project focuses on understanding the protein-protein interactions through detailed molecular simulations. Simulations are performed in explicit water; this results in systems that are computationally expensive to simulate through straightforward molecular dynamics. Researchers apply a recently developed advanced sampling technique called Forward Flux Sampling (FFS) in conjunction with MD simulations to overcome these limitations. FFS based calculations of protein complex formation leads to several hundred trajectories of molecular formations that successfully reached the next interface or molecular cluster size. Thousands of simulations must be run in order to find these 100+ successful trajectories. The mechanism of protein complex formation can be elucidated through the analysis of these trajectories. However, current tools do not allow for the visualizing the characteristics of these 100+ trajectories concurrently. While they can perform calculations based on these trajectories, the two key visualization challenges they face are (i) the molecular rendering of these trajectories simultaneously rather than one trajectory at a time, and (ii) visualizing the correlation between different parameters that may play a role in governing the interaction mechanism. I will work closely with the team of chemical engineers working on this project to address these challenges and develop visualization tools that will enhance our ability to view and analyze the results.

Research Purpose:

The purpose for this research experience is to determine the best way to help visualize pure water freezing rare event simulation data produced by the simulations run by Dr. Sarupria and her colleagues. Through background research, testing, and modification of visualization tools, I hope to reveal the best way to provide insight into Dr. Sarupria's research. I hypothesize that visualization techniques well equipped to depict multi-dimensional data in a clear, understandable fashion and tools that allow for manipulation and selection of the data will be most advantageous. It is important that this visualization can help find patterns within a large amount of data.

Week 1

Well, the program finally started, and I've been immersed in my first week of research experience this summer. I've started out by having to learn and understand a large amount of new information including Linux, Processing, and Java programming languages. Additionally, I have been introduced in more detail to some of the research being done in the chemical engineering department. This project is meant to simulate the rare event of absolutely pure water freezing in hopes that this technique of simulation and visualization could reveal optimal freezing cluster formations and lead to eventual simulations of protein folding. After some productive meetings this week, my mentor, Dr. Levine, introduced to my lab group and I to a visualization tool named Parallel Coordinates. After spending much time researching and studying background information on parallel coordinates, Dr. Levine introduced me to a parallel coordinates program named EDEN, produced by his friend Chad Steed. I spent some time playing around with this program and exploring all of its features. By the end of the week I had obtained some data for the project that I could visualize in EDEN.

More information on Parallel Coordinates!

More information about EDEN EDEN example

Week 2

Week 2 started off nicely with contuing learning and several informative lectures. At the start of this week, I was able to obtain the source code for EDEN from Chad Steed via Github. After talking with Dr. Levine and others in the chemical engineering lab, I shifted my focus to modifying the EDEN source code to include additional features that would be specifically useful for visualizing our project. The first task on my plate was to modify the code to include more variation for color selection. EDEN only contains coloring options for "selected" and "unselected" data and does not currently support multiple coloring of different selections. This task deemed itself more difficult than anticipated, but after some effort and time digging through the EDEN source code (written in Java, which I'm still learning) I managed to get EDEN to allow for multiple color selections.

Why coloring can be useful in Parallel Coordinates: Interactive example

EDEN example

In other news: Excited for the start for the World Cup!

EDEN example

USA win over Ghana 2-1!

EDEN example

Week 3

This week started with a little frustration as far as coding some new features for EDEN. I had originally planed to recolor the scatter plots in EDEN in conjunction with the lines, however, it seems that its not so easy to open multiple scatterplots while maintaining selection capability and allowing for recoloring. Dr. Levine and I agreed that I should table this goal for now. So, I moved on to creating an option on eden to color by a certain axis. What I mean by this is upon selecting "Color by Axis" and selecting a certain column an ordered colormap will be applied to all of the lines based on the order of the lines on the column you selected. The following pictures show what I'm trying to explain. As you can see, in these two examples, an HSV rainbow colormap is used. My hope is to eventually change the colormap to a directional colormap which would contain one color fading to another color with white in the middle (See Picture 3). Additionally, I figured out how to execute an external file from the EDEN Java code. This will be helpful for opening 3D representations of the ice clusters and displaying generated density maps based on selected lines in EDEN.

EDEN example EDEN example EDEN example

Good news! For the first time ever, the United States advances to the round of 16 for the 2nd straight world cup! We face Belgium next!

EDEN example

Week 4

This week I was able to update some of the color features seen below and also introduce a new zoom feature. I'm still having issues with a few bugs, but hopefully those will be dealt with soon.

This is an example of the updated color by axis feature, which now goes from blue to white to red. I also included in this picture an example of selection and the option to change the transparency of the unselected data.

EDEN example

This is an example of the "zoom" feature that I have added to EDEN. I changed the Prob_Cross axis scale from 0.0-0.87 to 0.5-0.87 by entering new values into the table. Doing so focuses in on a smaller region of the axis and removes any lines not within that new specified range.

EDEN example

Week 5

I didn't get too much done on my project this week due to preparations for our upcoming XSEDE conference in Atlanta, GA and our midterm presentations. However, I was able to fix some major bugs in the "zoom" feature. I hope to be able to work on expanding the file input system in the coming week and work on my poster for the conference.

Week 6

This week was mostly spent on making my poster for the XSEDE conference next week, but I was able to complete some goals for this week. I managed to fix some more bugs in the code and I was able to expand the file input system to allow for file names associated with each individual line. Next week we'll be in Atlanta, but I'm hoping to still get work done on the project.

Week 7

Most of this week was spent in Atlanta at the XSEDE conference where I presented a poster and attended many talks.

XSEDE Experience Report

Overall, this conference was a very good experience to develop my professional skills and gain networking experience. I felt most of the conference was focused on high performance computing, of which I have very little experience, and was only minimally focused on visualization. However, I do feel that it was a valuable experience to learn how a professional conference is conducted and how the environment of a conference feels.

Introduction to Modeling in Sage

This tutorial started with an introduction to the programming tool called sage. Sage is a computational mathematics tool that is very useful for quickly solving simple and complex mathematical systems. For example, you could solve a simple physics problem or you could solve a complex linear system. One important feature of Sage that is particularly useful for mathematics is the ability to perform calculations and solve systems while maintaining variables within the problem. This means that you don’t necessarily need to put any numbers into the problem to find a solution. Sage is also particularly useful for mathematical modeling using differential equations. Specifically, we looked at approximating equations using Euler’s method of approximation and the modified Euler’s method of approximation. This lecture really showed how Sage can be useful for modeling.

Can a $25+ Billion Investment Help California Balance Water Supply Reliability and Ecosystem Sustainability? Does Science Play an Important Part?

The format of this talk was interesting because it shifted back and forth between a politician and scientist speaking on the issue of water sustainability in California. I felt that both speakers brought up many points of why California needs sustainability and the difficulty to achieving sustainability. The contrast of the two speakers highlighted part of the difficulty of scientists and politicians working together.

Statistical Performance Analysis for Scientific Applications

The background of this lecture was interesting, but I found it difficult to follow along due to my unfamiliarity with high performance computing. The focus of this talk was optimizing the performance of a computer by statistically analyzing the outputs of a computer. Ideally, the best combination of outputs would increase the performance of the computer.

Revolutionizing the Next-generation Therapeutic Drug Discovery

The man giving this talk was working closely with a researcher at MUSC. The researcher at MUSC was involved in therapeutic drug discovery, but his algorithms for analysis were taking about 3 months of constant computing to complete 1 drug. At this rate, his research for 10 drugs was going to take over 2 years. The computer scientist giving this talk was able to optimize his algorithm to get the computing time for one drug down to about 40 minutes, which is an incredible increase in efficiency. I was immediately blown away when I heard how much the speaker was able to optimize the algorithm. I didn’t understand how he managed to do this, but it was pretty spectacular.

Graduate School Speed Networking

I attended this event hoping to meet some representative from graduate schools, but it turned out to be graduate students from various institutions. So, it was not exactly what I was expecting, but it was still useful. I managed to talk to quite a few graduate students, and even though none of them were in the field I’m interested in pursuing, all of them offered me valuable advice.

The Need for Computational Modeling Expertise in Industry

I found this lecture one of the most interesting and entertaining. The speaker was full of energy and the visuals on the slides captured your eyes. The lecture was about the uses of modeling and simulation in industry, specifically for Procter and Gamble. He introduced many uses for simulations that I would not have thought about. He also offered some comedic relief that really made this lecture worth the while.

Poster Showcase

I was able to present my own poster at this showcase for the second hour where I answered questions and explained my poster to several groups of people. During the first hour I managed to see some very impressive posters from graduate and undergraduate students. I also enjoyed the visualization showcase where researchers showed complex visualizations on large monitors related to their research.

Week 8

Now that we're in the final week of the program, I've been working on completing my work, tidying up my software, and making final presentations. However, I was still able to add some new features to EDEN. I managed to complete an addition to EDEN that allows for the user to open selected line simulation files in VMD (3D molecular model software). Additionally, I completed a a user documentation detailing how to use most of the features in the modified EDEN.


Pure water freezes spontaneously through forming ice clusters after reaching low temperatures. Normally, small impurities within water facilitate the freezing and can change how pure water freezes. There is currently no way to remove 100% of these impurities from water. Thus, determining how completely pure water freezes is difficult to study experimentally. Through molecular dynamics simulations combined with forward flux sampling (FFS) technique to overcome computationally expensive direct molecular dynamics simulations, it is possible to computationally imitate freezing of pure water. These computations produce large amounts data that make patterns difficult to interpret. Visualization of this data helps provide insights into patterns and how pure water freezes. Parallel coordinates is a useful visualization tool that allows for multidimensional comparison of different criteria calculated from the data. In this work, a parallel coordinates program named EDEN was modified to add several new features that enable enhanced analysis of particular simulated molecular formation elucidating how pure water freezes. Namely, these features are multiple color selection, color by axis, zoom, and opening simulation files. This expanded parallel coordinates tool is particularly useful because it allows for comparison of different measures within the data while also providing tools to analyze those specific comparisons. Ultimately, this tool will help determine optimal conditions that allow pure water to freeze with eventual hopes that it can be used to determine the effects of environmental conditions on phase transitions in aqueous systems.

Last updated: 7/24/2014