Value of SNPs With Very Small Effects in The Predictive Ability of Polygenic Risk Score: Illustrated Using Type II Diabetes Mellitus 公开

Ning, Kunru (Spring 2021)

Permanent URL: https://etd.library.emory.edu/concern/etds/v118rf581?locale=zh
Published

Abstract

Background: Polygenic risk score (PRS) is an index calculated by summing up multiplications of the number of risk alleles and effects of the single nucleotide polymorphism (SNP) to predict the risk of developing diseases for individuals. PRSs calculated by thousands and millions of SNPs with small effects have become a trend in many studies. We studied the value of including SNPs with very small effects when predicting disease risk using PRS, answering the following research questions: 1. Will the discriminative accuracy change when changing the precision of SNPs with small effects? 2. Will the predicted risks change when changing the precision of SNPs?

 

Methods: The analysis was conducted using simulated data. Hypothetical populations of 100,000 people, in which we predicted disease using 7502 SNPs associated with type II diabetes at population risk of 10% and 30%. A genetic profile with predicted disease risk and disease status was generated for each individual. AUC was calculated to quantify the predictive ability of the PRS models using SNPs with small effects that are non-zero in the 6th, 5th, 4th, 3rd, 2nd, and 1st decimal. Predicted risks were calculated 6 times, using SNP effects that are non-zero in the 6 decimal levels. Correlation coefficients were obtained for the predicted risks to measure the strength of association.

 

Results: When the SNP effects were kept in the 5th, 4th, and 3rd decimals, the predicted ability of PRS remained the same (AUC = 0.65, correlation coefficient = 1.0) comparing with when the SNP effects were kept in the 6th decimal, for both population risk of 10% and 30%. When SNP effects were kept in the 2nd decimal, the AUC remained the same at 0.65 with a slightly lower correlation coefficient of 0.97. The AUC and correlation coefficient reduced to 0.64 and 0.74 when SNP effects kept in the 1st decimal.

 

Conclusion: The study concluded that including SNPs with very small effects in the predictive ability of PRS did not change predicted risks. Therefore, there is no necessity of including SNPs with very small effects when predicting PRS in future studies, keeping SNP effects in the 2nd decimal with a precision of 0.01 should suffice.

Table of Contents

Table of Contents

INTRODUCTION............................................................................................................................- 1 - METHOD...........................................................................................................................................- 4 -

Data Collection..............................................................................................................................- 4 -

Data Preparation ...........................................................................................................................- 4 -

Simulating Population..................................................................................................................- 4 -

Calculating AUC...........................................................................................................................- 5 -

Calculate predicted risk ...............................................................................................................- 5 -

Statistical Analysis ........................................................................................................................- 5 -

RESULTS ...........................................................................................................................................- 7 -

DISCUSSION .................................................................................................................................- 14 -

Strength and Limitations ...........................................................................................................- 15 -

Conclusion and Implications.....................................................................................................- 15 - REFERENCE....................................................................................................................................- 16 -

About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
School
Department
Subfield / Discipline
Degree
Submission
Language
  • English
Research Field
关键词
Committee Chair / Thesis Advisor
最新修改

Primary PDF

Supplemental Files