NIAPU: network-informed adaptive positive-unlabeled learning for disease gene identification.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Understanding gene-disease associations is crucial for disease etiology and treatment development.
• Identifying novel disease-associated genes is challenging due to limited studies.
• Prior knowledge and computational approaches like positive-unlabeled learning can aid gene discovery.

Purpose of the Study:

• To propose a novel Markov diffusion-based multi-class labeling strategy for putative disease gene discovery.
• To introduce effective network-based features for prioritizing candidate disease genes.
• To enhance computational search for new disease genes using machine learning.

Main Methods:

• Developed a novel Markov diffusion-based multi-class labeling strategy.
• Proposed a set of effective network-based features for gene prioritization.
• Utilized positive-unlabeled learning within a machine learning framework.

Main Results:

• The new labeling algorithm and features were tested on 10 disease datasets using three machine learning algorithms.
• Proposed features demonstrated competitive predictive power against state-of-the-art methods.
• The integrated methodology proved effective in searching for novel disease genes.

Conclusions:

• The proposed network-based features and Markov diffusion strategy offer a competitive approach for putative disease gene discovery.
• This methodology can significantly ease the computational search for new candidate disease genes.
• The findings contribute to a better understanding of gene-disease associations and potential therapeutic targets.