Bram Stone

**** ****** ** *** ** Oxford, MS 662-***-****

ac4ttv@r.postjobfree.com https://bramstone.github.io https://linkedin.com/in/bramstone Biologist and data enthusiast with signi cant experience in ETL processes, statistical modeling, and Skills data-driven discovery in the plant microbiome. Scienti c Research presentation, experimental design, bioinformatics, microbiology, DNA extraction, PCR, microbial culturing (agar plating and broth), uorometry Statistical

Experience

linear mixed models, non-linear mixed models multivariate statistics, network analyses, metagenomic analyses, machine learning

Software

R (dplyr, data.table, ggplot2, caret, igraph, text2vec, parallel, Rmarkdown), Python (pandas, numpy, matplotlib, biopython, Jupyter Notebooks), MySQL, mothur, BLASTx tools, bash scripting

Education

Ph.D. University of Mississippi Biology 2018

B.S. Portland State University Environmental Science and Management 2011 Experience

University of Mississippi August 2013 – current

Graduate • Analyzed Analyst the causitive / Biologist factors of seasonal shifts in the plant microbiome

• Predicted changes in bacterial diversity using linear regression and AIC for model selection

• Authored four peer-reviewed publications and presented at four scienti c conferences

• Increased retention of DNA material in lab procedures 12.5% by analyzing DNA sequence data leading to 4% reduction in per project expenses

• Managed and mentored three undergraduate workers involved in research activities Research Projects

Machine • Constructed Learning random Regression forest regression of bacterial diversity across time in response to climate

• Improved R2 by 18% compared to out-of-bag model by applying custom feature engineering function to calculate rolling summaries

• Produced new insight into system by identifying rain and wind as major drivers of microbiome Biological • Built bootstrap Data Mining algorithm and to Network mine signi cant Analysis biological interactions in highly correlated dataset

• Achieved minimal false positive error rate (0.003% at 95% con dence) while performing feature selection across 450 million pairwise feature associations

• Discovered key network species for further experimentation using degree centrality