Estimating mutual information using B-spline functions--an improved similarity measure for analysing gene expression...

Area of Science:

• Bioinformatics
• Computational Biology
• Information Theory

Background:

• Mutual information is a key concept in information theory for assessing variable dependencies.
• It's proposed as a similarity measure for gene expression clustering, extending linear methods.
• Standard application to continuous data requires binning, risking numerical errors in smaller datasets.

Purpose of the Study:

• To develop a numerical method for estimating mutual information from continuous data.
• To enhance the accuracy and significance of mutual information as a similarity measure.
• To improve the detection of non-linear correlations in gene expression datasets.

Main Methods:

• A novel algorithm for the numerical estimation of mutual information from continuous data is proposed.
• The algorithm's properties are investigated, and its performance is compared to existing methods.
• The method is applied to large-scale gene expression datasets.

Main Results:

• The proposed method significantly increases the power of distinction from random correlation.
• It outperforms commonly used algorithms in estimating mutual information from continuous data.
• Analysis on gene expression datasets demonstrates improved detection of non-linear relationships.

Conclusions:

• Mutual information serves as a powerful similarity measure for uncovering non-linear correlations in gene expression data.
• This approach extends the utility of linear correlation measures, offering a more comprehensive analysis.
• The developed algorithm provides a robust tool for bioinformatics research.