Myers Laboratory (PI: Chad Myers, PhD)- Developing predictive models of cancer genetic dependencies using data from the Broad Institute’s cancer dependency map project (DepMap).- Conducted a re-analysis of a previously published deep learning method for predicting cancergenetic dependencies, concluding that the method fails to outperform simple baseline models (Chang et al, submitted).- Exploring the use of network smoothing and utilizing gene expression prediction from mutationprofiles as a pre-training task to better represent functional effects of cancer mutation profiles.

Microbiome Systems Medicine Laboratory (PI: Jaeyun Sung, PhD)- Developed GMWI-webtool, a browser application for computing the lab’s previously publishedGut Microbiome Wellness Index (GMWI), a gut microbiome based health status predictor (Chang et al, Bioinformatics 2023).GMWI-webtool was used by the lab to apply GMWI as a quantitative tool for measuring theprebiotic effect of oligosaccharides on the gut microbiome (Lee*, Seong*, Chang et al, npj Science of Food 2023).- Conducted a pooled analysis of 8,069 stool metagenomes from 54 independent studies across 12health and disease phenotypes to develop Gut Microbiome Wellness Index 2 (GMWI2). UsingLasso-penalized logistic regression, GMWI2 achieved a cross-validation balanced accuracy of80% in predicting disease presence from gut microbiome taxonomic profiles (nearly 10% higherthan that of its predecessor) (Chang* and Gupta* et al, Nature Communications 2024).- Developed software for the lab’s TaxiBGC pipeline, a computational tool for detecting biosynthetic gene clusters in metagenomes by first identifying the microbes likely to harbor them (Gupta, Bakshi, Chang et al, mSystems 2022).- Proposed a computational framework for investigating the role of Human Endogenous Retroviruses (HERVs) in human disease. Used structural bioinformatics tools to identify HERV geneproducts that potentially induce autoimmunity via molecular mimicry.

Knights Laboratory (PI: Dan Knights, PhD)- Analyzed the lab’s Microbiome, Antibiotics, and Growth Infant Cohort (MAGIC), a longitudinalbirth cohort of 386 children. Used computational techniques to dissect the structure and functionof the developing gut microbiome.- Used multivariate ecology methods to observe long term species-level taxonomic compositioneffects of perinatal antibiotic exposure.- Inferred microbial growth rates by computing peak-to-trough ratios (PTR) of genome readcoverage, identifying microbes with significantly different growth rates between infants with andwithout perinatal antibiotic exposure, and discovering that microbes rapidly colonize the gut inthe first few weeks after birth followed by stagnation.- Developed a reference genome- and assembly-free computational framework for discovering novelbiological entities and ecological relationships in metagenomic data de novo.Minnesota Supercomputing Institute Research Exhibition Poster Award - $1,000

◦ Co-founded the first youth UNICEF chapter in MN◦ Networked with local businesses to raise $12,000+ for disaster relief and with local schools for advocacy events.◦ Created excel scripts for managing fundraising event budgets and managed monetary transactions with UNICEF.

◦ Performed electrophysiology trials on functional defects of Dent disease causing mutations to quantify ion exchange◦ Cloned wild type CLC-5 proteins in Xenopus laevis oocytes and performed eGFP and hemagglutinin tagging to quantify trafficking defects.