home arrow-down arrow-up menu new-tab phone fax mail

University of Hawaiʻi at Hilo

ʻIke Wai Scholars Projects

The application period for the ‘Ike Wai Scholars program is now closed. We expect to run this program again in the 2018-2019 school year. Mahalo for your interest in our program

2017-2018 ʻIke Wai Scholars Projects

UH-Hilo is pleased to offer paid ($500 monthly) internships to undergraduates with interests in Data Science. As an emerging field, Data Science uses techniques and advancements in Statistics, Modeling, Computer Science, and various scientific fields to collect, analyze, and visualize large and complicated data sets.

Six ‘Ike Wai Scholars will be expected to work an average of 40 hours per month, each with a UH-Hilo research mentor. These projects range in topics from Machine Learning to Bioinformatics. Project descriptions can be found below.

Please contact Dr. Bob Pelayo (robertop@hawaii.edu) for more information on the ‘Ike Wai Scholars program.

Project 1: Exploring Collections of Phylogenetic Trees

Faculty Mentor: Dr. Grady Weyenberg, Mathematics

Project Description: Phylogenetic trees are used to represent the evolutionary relationships between different organisms. Modern sequencing methods have enabled biologists to collect large amounts of sequence data across the genomes of many organisms, but for various reasons, creating a phylogenetic tree using whole-genome data does not produce a better understanding of the evolutionary relationships. Instead of building a single tree using all of the data, one should create many different trees from separate genomic regions. However, this leads to difficulties, because it is to synthesize information from many trees into a single conclusion.

In this project, the student will learn the basics of how to construct phylogenetic trees from sequence data, and learn techniques for comparing different trees to each other. We will attempt to better understand the statistical properties of some new methods of summarizing information from sets of trees. If possible, we will also look at the sequences of some native Hawaiian plants and attempt to better understand their evolutionary relationships.

Desired Skills: The ideal candidate for this project will have some knowledge of the basics of molecular evolution and biological sequence data, as well as some experience with command-line computing tools, and at least a small amount of statistics or programming background knowledge.

Project 2: 3D Coral Reef Data Visualization Project

Faculty Mentor: Dr. John Burns, Marine Science Contact Info: johnhr@hawaii.edu, office: Wentworth 2

Project Description: The 3D coral reef data visualization project will analyze 3D reconstructions of Hawaiian reefs derived from cutting edge structure-from-motion photogrammetry. The scholar will be involved in creating 3D reef models, annotating models, and analyzing structural data to assess habitat dynamics of coral reef environments. The scholar will also work to develop new automated annotation techniques using several software programs. The goal of the automation process will be to create a data pipeline to automatically extract spatial and structural information of coral communities from the 3D models.

Desired Skills: This project will involve using several computer vision programs, and will require a detail-oriented approach. The scholar will learn the full process of 3D data visualization from image acquisition in the field to data processing and statistical analysis. Basic skills in computer rendering, ArcGIS, and statistics are preferable.

Project 3: What’s in the Local Water and Soil?

Faculty Mentors: Dr. Natalie Crist and Matthew Platz, Chemistry

Project Description: Big Island agriculture utilizes fertilizer and other chemicals to control pests and weeds. Do these chemicals migrate into local waters? Human and animal waste contains chemicals, especially high levels of nitrogen and phosphorus. Do they migrate into local waters?

The Big Island once was home to large sugar plantations. Phosphorus levels in the soils were measured thirty years ago, at the height of sugar production. Do abandoned sugar plantations still have similar amounts of nitrogen and phosphorus in the soil?

In this project an ‘Ike Wai Scholar will work with Chemistry faculty to determine nitrogen, phosphorus, pesticide and herbicide levels in water and soil samples collected around the Big Island of Hawai’i.

Desired Skills: It is preferred that the “Ike Wai Scholar be a Chemistry major and have completed Chemistry 161-2 and 241-2 and the laboratory classes associated with these courses, and have a GPA in Chemistry above 3.0. The ‘Ike Wai Scholar will learn advanced analytical chemistry techniques and gain experience using modern instrumentation. Additionally, they will have the opportunity to visit various interesting agricultural sites around the Big Island of Hawai’i.

Project 4: Effects of Rare Variants in Concurrent Drug Usage

Faculty Mentor: Dr. Michael R. Peterson, Computer Science

Project Description: In the present age of data collection, there is a vast store of unexamined or insufficiently examined medical and pharmacology data that, if properly and thoroughly analyzed, could reveal many more important, life-saving revelations. In particular, in pharmacology, there are gigantic untapped databases, which have only been examined with the bare minimum of techniques and research questions. Thus, we propose to ask a new research question in the area of analyzing pharmacology datasets and borrow techniques from genome wide association studies (GWAS) in an attempt to answer the question, “Do rare variants in concurrent drug usage effect end outcomes?” The influence of rare variants in the concurrent drug usage on the outcome of interest has not been analyzed in-depth. The standard method of analysis has only encompassed testing of the significance of common variants in concurrent drug usage, comorbidities, and genetic markers. This study proposes the application of state of the art association analysis tools: Combined Multivariate Collapsing (CMC), Kernel-Based Adaptive Clustering (K-BAC), and Weighted Sum Statistics (WSS) algorithms, for testing association of patient drug usage description data to drug treatment outcome. It is also interested in the modification of these methods for the specific purpose of analyzing pharmacology data with higher sensitivity and specificity.

We hope to demonstrate the usefulness of this approach in detecting significant rare descriptive data sequences and perform an analysis of a pharmacology dataset, featuring patients treated with Warfarin. We will examine the association of comorbidities and concurrent drug usage with reaching stable dosage levels of Warfarin. Statistically significant results will reveal important, undocumented associations between concurrent drug usage and a patient’s ability to reach stable dosage levels of Warfarin. We consider rare variants in self-governed medical decisions, which may lead to significant difference in dosage stability outcomes. Within the data analysis of the Warfarin dataset, we wish to reveal statistically significant results in rare variants of concurrent drug usage which provide higher probability of either reaching or not reaching stable dosage levels of Warfarin. The goal of the project, however, is to further develop, rigorously test, and justify the application of CMC, K-BAC, and WSS algorithms for the analysis of rare variants in pharmacology datasets.

Desired Skills: A solid background in statistics, R, GWAS, and linear algebra will ease the practicalities of this project. Project deliverables include a conference and/or journal manuscript, as appropriate.

Project 5: Beak diversity in birds: which genes fit the bill?

Faculty Mentor: Dr. Mona Renee R Bellinger, Biology

Project Description: Hawaiian honeycreepers diversified from an ancestral species of finch that colonized the Hawaiian archipelago somewhere between 20 to 5 million years ago. This group of birds exhibits tremendous morphological diversity, with bill shapes and sizes that correspond to feeding specialization (Figure 1). Candidate genes believed to underpin beak features have been identified from other finch species, but these genes have not yet been examined in Hawaiian honeycreepers. The goal of this project is to mine through existing Hawaiian Honeycreeper genomic resources to determine if patterns across candidate genes can explain the bill diversity we observe in Hawaiian birds. The three objectives of this study are to: (1) compile a list of candidate genes related to bill morphology; (2) develop a representative dataset of candidate genes by data mining bird genomes (>100 are available); and (3) mine Hawaiian Honeycreeper sequence datasets to assess if nucleotide diversity at candidate genes shows correspondence to bill shape and size. Depending on student interest and availability of funds, we can develop primer pairs for a sub-set of “best performing” genes to amplify and sequence them across ~5 species of Hawaiian Honeycreepers. This would permit testing individual candidate gene hypotheses.

Desired Skills: The ideal student for this opportunity will have strong interest in avian biology and genomics, willingness to learn shell scripting and to work off of a supercomputer, willingness to learn genetic software packages and tools for mining genomes, have strong attention to detail, and ability to maintain careful notes, including a laboratory notebook. No programming experience necessary.

Figure 1: Illustration of Hawaiian Honeycreepers that radiated in the Hawaiian archipelago (by Douglas Pratt). Figure 1: Illustration of Hawaiian Honeycreepers that radiated in the Hawaiian archipelago (by Douglas Pratt).

Project 6: Optimizing Human-in-the-Loop Artificial Intelligence Though Analysis of Video Game Data

Faculty Mentor: Dr. Travis Mandel, Computer Science

Project Description: Artificial Intelligence is rapidly gaining momentum, with AI systems conquering board games, controlling robots, and even driving cars. These successes have been largely driven by the ease of collecting vast quantities of high-quality data. Yet developing AI systems that design video games, develop treatment regimes, or create personalized lesson plans is in many ways a much more challenging problem due to the difficulty of collecting these huge datasets. To help achieve good performance with limited data, we will investigate how to best include humans “in the loop”, leveraging their intuition and creativity to rapidly improve these AI systems.

In our first project, we will examine student data gathered from a large-scale live experiment on an educational video game. Your task will be to investigate the best way for non-expert human workers (gathered from online platforms like mechanical turk) to improve the system, for example by relaxing constraints and thereby broadening the set of interventions available. Such a task is highly nontrivial due to the tendency of these workers to misunderstand the task or game the system. Yet achieving this could potentially have a large positive impact on student learning while significantly reducing the amount of effort required by education experts.

As the academic year progresses, additional opportunities will arise, including opportunities to work on data from different games and explore various other interesting problems in Human-in-the-Loop AI.

Desired Skills: At a minimum, applicants should have successfully completed CS 321.