CoRSIV-based model, with a positive predictive value of 80 percent, surpassed the performance of a model based on polygenic risk score
MONDAY, Aug. 9, 2021 (HealthDay News) — A machine learning case-control classifier can identify schizophrenia based on DNA methylation in the blood, according to a study published online Aug. 3 in Translational Psychiatry.
Chathura J. Gunasekara, Ph.D., from the Baylor College of Medicine in Houston, and colleagues trained a schizophrenia case-control classifier based on DNA methylation in the blood, focusing on human genomic regions of systemic interindividual epigenetic variation (CoRSIVs); a subset of these are represented on the Human Methylation 450K (HM450) array. HM450 DNA methylation data from 414 schizophrenia cases and 433 nonpsychiatric controls were used as training data for a classification algorithm. A “risk distance” was calculated to identify individuals with the highest probability of schizophrenia using the first two sparse partial least squares discriminate analysis dimensions. The model was then evaluated on 353 schizophrenia cases and 322 nonpsychiatric controls.
The researchers found that the CoRSIV-based model classified 303 individuals as cases, with a positive predictive value of 80 percent, which surpassed performance of a polygenic risk score-based model. Risk distance based on methylation was not associated with medication use. There was a positive correlation observed for risk distance and polygenic risk score; a mediational analysis indicated that genetic effects were partly mediated by altered methylation at CoRSIVs.
“We consider our study a proof of principle that focusing on CoRSIVs makes epigenetic epidemiology possible,” a coauthor said in a statement.
Copyright © 2021 HealthDay. All rights reserved.