Prior information-assisted integrative analysis of multiple datasets

Area of Science:

• Biomedical research
• Genetics
• Computational biology

Background:

• Genetic studies face challenges with small sample sizes and high dimensionality, limiting predictive model accuracy.
• Current integrative analysis methods often lack efficient variable selection strategies.
• Penalization techniques are used but can be inefficient when searching vast variable spaces.

Purpose of the Study:

• To develop a method for incorporating prior biological information into the integrative analysis of multiple genetic datasets.
• To improve the efficiency and accuracy of identifying genetic markers and constructing predictive models.
• To address the limitations of small sample size and high dimensionality in genetic research.

Main Methods:

• Utilized a convolutional neural network with active learning to extract prior information from textual data of previous studies.
• Incorporated extracted prior information into a group LASSO-based technique for integrative analysis.
• Validated the method through simulation studies and analysis of skin cutaneous melanoma data.

Main Results:

• The proposed method demonstrated satisfactory performance in simulation studies.
• Successfully applied the method to analyze skin cutaneous melanoma data, establishing practical utility.
• The integration of prior information enhanced the analysis of multiple genetic datasets.

Conclusions:

• The developed method effectively incorporates prior information to improve integrative genetic data analysis.
• This approach offers a more efficient and accurate way to handle the 'small sample size, high dimensionality' problem in genetics.
• The findings have practical implications for biomedical research, particularly in cancer genomics.