NetMoST: A network-based machine learning approach for subtyping schizophrenia using polygenic SNP allele biomarkers

Area of Science:

• Neuroscience
• Genetics
• Computational Biology

Background:

• Schizophrenia subtyping is crucial for effective diagnosis and treatment of this complex, polygenic disorder.
• Current symptom-based diagnosis for schizophrenia is often unreliable due to genetic heterogeneity.
• Novel methods are needed to overcome limitations in subtyping complex psychiatric disorders.

Approach:

• Developed netMoST, a network-based machine learning approach for psychiatric disorder subtyping.
• Utilized genome-wide genotyping data to identify polygenic risk SNP-allele modules as polygenic haplotype biomarkers (PHBs).
• Applied netMoST to a schizophrenia cohort, identifying three distinct biotypes.

Key Points:

• The three schizophrenia biotypes exhibited differentiable genetic, neuroimaging, and functional characteristics.
• Biotype 1 (36.9%) PHBs related to neurodevelopment and cognition.
• Biotype 2 (28.4%) PHBs enriched for neuroimmune functions.
• Biotype 3 (34.7%) PHBs associated with calcium ion and neurotransmitter transport.
• Neuroimaging (regional homogeneity) patterns supported the distinct biotypes compared to controls.

Conclusions:

• netMoST effectively uncovers novel biotypes in complex diseases like schizophrenia.
• The study highlights the utility of exploring polygenic allelic patterns beyond conventional Genome-Wide Association Studies (GWAS).
• Findings pave the way for more precise diagnostic and therapeutic strategies in psychiatry.