Sequencing-based studies have recognized hundreds of genetic variants that increase the risk of schizophrenia (SCZ), but only a few percents of heritability can be attributed to these loci. It is challenging to discover the full spectrum of schizophrenia genes and reveal the dysregulated functions underlying the disease. Here, we proposed a holistic model for predicting disease genes (HMPDG), a novel machine learning prediction strategy integrated by Protein-Protein Interaction Network (PPIN), pathogenicity score, and RNA expression data. Applying HMPDG, 1946 potential risk genes (PRGs) as a complement of the genetic basis of SCZ were predicted. Among these, the first decile genes were highlighted as high confidence genes (HCGs). PRGs were validated by multiple independent studies of schizophrenia, including genome-wide association studies (GWASs), gene expression studies, and epigenetic studies. Remarkably, the strategy revealed causal genes of schizophrenia in GWAS loci and regions of copy number variant (CNV), providing a new insight to identify key genes in disease-related loci with multi genes. Leveraging our predictions, we depict the spatiotemporal expression pattern and functional groups of schizophrenia risk genes, which can help us figure out the pathophysiology of schizophrenia and facilitate the discovery of biomarkers. Taken together, our strategy will advance the understanding of schizophrenia genetic basis and the development of diagnosis and therapeutics.

Download data

• Downloaded 237 times
• Download rankings, all-time:
  • Site-wide: 93,387
  • In psychiatry and clinical psychology: 331
• Year to date:
  • Site-wide: 38,585
• Since beginning of last month:
  • Site-wide: 38,585

Altmetric data

Downloads over time

Distribution of downloads per paper, site-wide