Complex Disease Genes Identification Using a Heterogeneous Network Embedding Approach.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Experimental methods for gene-disease association are costly and time-consuming.
• Computational approaches offer a cost-efficient alternative for identifying candidate genes.
• Accurate identification of disease-associated genes is crucial for understanding complex diseases.

Purpose of the Study:

• To propose NetEM, a novel heterogeneous biological network embedding approach for identifying disease-associated genes.
• To evaluate the performance of NetEM against existing state-of-the-art algorithms.
• To validate NetEM using real-world clinical data.

Main Methods:

• Developed NetEM, a heterogeneous biological network embedding approach.
• Evaluated NetEM on six complex diseases: peroxisomal disorders, sarcoma, Grave's disease, lysosomal storage diseases, blood coagulation disorders, and hypertrophic cardiomyopathy.
• Validated NetEM using TCGA data for Invasive Lobular Breast Cancer and CPTAC data for human glioblastoma.

Main Results:

• NetEM demonstrated superior performance in identifying disease genes compared to Cardigan, DIAMOnD, and GeneWanderer.
• Case studies using TCGA and CPTAC data confirmed the validity and effectiveness of the NetEM method.
• The proposed NetEM approach significantly improves the accuracy of disease-gene association identification.

Conclusions:

• NetEM provides a powerful and efficient computational tool for discovering disease-associated genes.
• The method's performance surpasses current state-of-the-art algorithms, offering a valuable resource for genetic research.
• NetEM's applicability to real clinical data underscores its potential impact on disease research and diagnostics.