Keratoconus-susceptibility gene identification by corneal thickness genome-wide association study and artificial intelligence IBM Watson

• Published in: Communications biology Vol. 3; no. 1; p. 410
• Main Authors: Hosoda, Yoshikatsu, Miyake, Masahiro, Meguro, Akira, Tabara, Yasuharu, Iwai, Sachiko, Ueda-Arakawa, Naoko, Nakano, Eri, Mori, Yuki, Yoshikawa, Munemitsu, Nakanishi, Hideo, Khor, Chiea-Chuen, Saw, Seang-Mei, Yamada, Ryo, Matsuda, Fumihiko, Cheng, Ching-Yu, Mizuki, Nobuhisa, Tsujikawa, Akitaka, Yamashiro, Kenji
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 31.07.2020; Nature Publishing Group
• Subjects: 631/114/2398; 631/208/205/2138; 692/699/3161/3163; Artificial intelligence; Basal cells; Biology; Biomedical and Life Sciences; Cornea; Corneal transplantation; Genome-wide association studies; Genomes; Keratoconus; Life Sciences; Smad3 protein; Susceptibility
• ISSN: 2399-3642; 2399-3642
• DOI: 10.1038/s42003-020-01137-3

Keratoconus is a common ocular disorder that causes progressive corneal thinning and is the leading indication for corneal transplantation. Central corneal thickness (CCT) is a highly heritable characteristic that is associated with keratoconus. In this two-stage genome-wide association study (GWAS) of CCT, we identified a locus for CCT, namely STON2 rs2371597 ( P  = 2.32 × 10 −13 ), and confirmed a significant association between STON2 rs2371597 and keratoconus development ( P  = 0.041). Additionally, strong STON2 expression was observed in mouse corneal epithelial basal cells. We also identified SMAD3 rs12913547 as a susceptibility locus for keratoconus development using predictive analysis with IBM’s Watson question answering computer system ( P  = 0.001). Further GWAS analyses combined with Watson could effectively reveal detailed pathways underlying keratoconus development. Yoshikatsu Hosoda et al. study the genetic basis for central corneal thickness (CCT) that is associated with keratoconus. They identify two susceptibility loci, STON2 rs2371597 and SMAD3 rs12913547, using two-step genome-wide association study (GWAS) and predictive analysis with IBM’s Watson question answering computer system, respectively.