DeepGP: An Integrated Deep Learning Method for Endocrine Disease Gene Prediction Using Omics Data.

Area of Science:

• Endocrinology and genetics
• Computational biology and bioinformatics

Background:

• The endocrine system regulates vital functions including metabolism, growth, and reproduction.
• Endocrine diseases, such as diabetes mellitus and polycystic ovary syndrome, significantly impact human health.
• Understanding the genetic basis of endocrine diseases is crucial for developing effective treatments.

Purpose of the Study:

• To develop a deep learning model for prioritizing genes associated with endocrine diseases.
• To identify susceptible genes for five common endocrine disorders using a novel computational approach.

Main Methods:

• A deep learning method, DeepGP, was developed, integrating graph convolutional networks and convolutional neural networks.
• The model was trained and validated on an integrated dataset using 10-cross-validations.
• Performance was evaluated using the area under the curve (AUC) and area under the precision-recall curve (AUPRC).

Main Results:

• DeepGP achieved an AUC of approximately 83% and an AUPRC of approximately 65% in prioritizing susceptible genes.
• The study found significant overlap in associated genes between type 1 diabetes mellitus (T1DM) and type 2 diabetes mellitus (T2DM).

Conclusions:

• DeepGP demonstrates effectiveness in identifying disease-associated genes for endocrine disorders.
• Further investigation into T1DM- and T2DM-specific genes is warranted for a deeper understanding of their pathogenesis.
• This approach can aid in discovering novel pathogenic genes and understanding endocrine disease mechanisms.