AUTHOR=Zhou Xueping , Zhang Jipeng , Ding Ying , Huang Heng , Li Yanming , Chen Wei 

TITLE=Predicting late-stage age-related macular degeneration by integrating marginally weak SNPs in GWA studies

JOURNAL=Frontiers in Genetics

VOLUME=Volume 14 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2023.1075824

DOI=10.3389/fgene.2023.1075824

ISSN=1664-8021

ABSTRACT=Introduction:  Age-related  macular  degeneration  (AMD)  is  a progressive  neurodegenerative disease and the leading cause of blindness in developed countries. Current late-stage AMD susceptible Single-Nucleotide Polymorphisms (SNPs) detection are mainly via single-marker based approaches, which investigate one marker  at a time and postpone integrating inter-marker connection information in the downstream fine mappings. Recent studies showed that directly incorporating inter-marker connections into genetic biomarker detection can discover novel marginally weak biomarkers often missed in marginal approaches, and at the same time, can significantly improve the disease prediction accuracy. 

Methods: Single-marker analysis is performed to detect marginally strong biomarkers using SNP genotype data. Linkage-disequilibrium is used to search SNP sets each containing SNPs connected to some marginally strong SNPs. Marginally weak SNPs are selected via a joint analysis upon all detected connected SNP sets. Prediction is made based on all selected marginally strong and weak signals.

Results: Previously found gene regions associated with late-stage AMD (e.g., BTBD16, C3, CFH, CFHR3, HTARA1) are confirmed. Novel genes DENND1B, PLK5, ARHGAP45, and BAG6  are discovered as marginally weak signals by integrating high linkage disequilibrium connections between SNPs. An overall prediction accuracy of 76.8% and 73.2% were achieved by with and without the identified marginally weak signals incorporated, respectively. 

Conclusions: Many marginally weak signals, raised from incorporating inter-marker connections, have strong predictive effects on AMD. Detecting and integrating such marginally weak signals can help identify new informative biomarkers and improve prediction performance.