fairfolio
Saved Projects

Using Machine Learning Algorithms to Investigate Genetic Burden of Scoliosis from Candidate Genes & GWAS Studies

CSEF · 2018 Computational Systems & Analysis

Elleen Kim · Palos Verdes Peninsula High School, Rolling Hills Estates, Los Angeles County

Overview

Objectives/Goals The engineering goal was to create programs that cluster characteristics of Scoliosis candidate genes and prioritize pairs of Scoliosis patient DNA sequences to investigate the causes of Scoliosis. Methods/Materials A database of candidate genes was compiled via literature review of published papers. Candidate genes were grouped by running their numerical characteristics through KNN and K-Means clustering algorithms. Scoliosis patient DNA sequences from public databases were run through customized local and global alignment programs and through BLAST. Results Results from the KNN program provided an upper bound and lower bound threshold to quantify the validation test scores of clustering, allowing patient data to be quantified by quality according to this threshold. KNN and K-Means are meaningful analyses of gene groups that can elucidate underlying similarities of Scoliosis candidate genes while producing consistent results within the margin of error. Results from alignment provided a prioritized list of Scoliosis DNA sequences. Conclusions/Discussion In conclusion, this project is a proof of concept experiment that shows that machine learning algorithms can be applied to genetic data. Prioritized DNA sequences can be used to further study the association to Scoliosis. With the falling cost and rising prevalence of GWAS studies, this prioritization can be used to quantify the influx of GWAS data for various diseases. Alignment programs also provided meaningful differences in best alignment scores between patient DNA sequences to use as a starting point for further research on what possible genetic variations or phenotypic characteristics determine best alignment. The machine learning clustering algorithms produced a range in validation test measures to quantify genetic datasets. Future identified candidate can be assessed using this metric. In line with our experimental goals, identified clusters or groups of candidate genes also serve as a foothold for further investigation on potential relationships or patterns between the genes and Scoliosis.

Summary statement

This project customizes machine learning clustering algorithms and DNA alignment algorithms to study Scoliosis candidate genes and patient DNA, and creates a metric to quantify the quality of GWAS datasets in other disease applications.

Help received

I designed and built the programs by myself. I got help in understanding the algorithms from Dr. Neil Sarkar (Department of Biomedical Informatics, Brown University) and Sophie Kim (Department of Humanities and Sciences, Stanford University).

Competition history

  • CSEF 2018 Computational Systems & Analysis · Entry S0815

Resources

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

Source: California Science & Engineering Fair public projects

Save projects to your library

Sign in with Google to keep track of projects you find interesting, organized into folders. An account also raises your daily allowance for “Has this been done?”, and lets you create a key for the MCP server with a much higher limit than anonymous use. Browsing stays public.

Continue with Google